|Home | About | Journals | Submit | Contact Us | Français|
Connexin-43 (Cx43), a predominant cardiac connexin, forms gap junctions (GJs) that facilitate electrical cell–cell coupling and unapposed/nonjunctional hemichannels that provide a pathway for the exchange of ions and metabolites between cytoplasm and extracellular milieu. Uncontrolled opening of hemichannels in the plasma membrane may be deleterious for the myocardium and blocking hemichannels may confer cardioprotection by preventing ionic imbalance, cell swelling and loss of critical metabolites. Currently, all known hemichannel inhibitors also block GJ channels, thereby disturbing electrical cell–cell communication. Here we aimed to characterize a nonapeptide, called Gap19, derived from the cytoplasmic loop (CL) of Cx43 as a hemichannel blocker and examined its effect on hemichannel currents in cardiomyocytes and its influence in cardiac outcome after ischemia/reperfusion. We report that Gap 19 inhibits Cx43 hemichannels without blocking GJ channels or Cx40/pannexin-1 hemichannels. Hemichannel inhibition is due to the binding of Gap19 to the C-terminus (CT) thereby preventing intramolecular CT–CL interactions. The peptide inhibited Cx43 hemichannel unitary currents in both HeLa cells exogenously expressing Cx43 and acutely isolated pig ventricular cardiomyocytes. Treatment with Gap19 prevented metabolic inhibition-enhanced hemichannel openings, protected cardiomyocytes against volume overload and cell death following ischemia/reperfusion in vitro and modestly decreased the infarct size after myocardial ischemia/reperfusion in mice in vivo. We conclude that preventing Cx43 hemichannel opening with Gap19 confers limited protective effects against myocardial ischemia/reperfusion injury.
Gap junction (GJ) channels are essential for the function of the heart and blood vessels by providing electrical coupling and direct cell–cell transfer of chemical/metabolic signals [20, 32, 45, 74]. They are composed of two docked hemichannels (connexons) oligomerized from six connexin molecules. The 43-kDa connexin protein (Cx43) is a major connexin in the heart and is especially abundant in ventricular cardiomyocytes . Aging and cardiac disease are associated with alterations in Cx43 expression, its localization and its phosphorylation status, and changes of GJ properties that, collectively, are thought to contribute to myocardial infarction injury and arrhythmogenesis [35, 64, 71]. Beyond GJs, emerging evidence has suggested novel roles of Cx43 hemichannels in the diseased myocardium. These unapposed/nonjunctional hemichannels reside in the zone surrounding the GJ nexus area called the perinexus . They are typically closed under normal conditions, but may open in response to ischemic insults resulting in ATP leakage, excessive entry of Na+ and Ca2+ and the loss of essential metabolites from the cells . Uncontrolled activation of hemichannels may potentially introduce significant changes in cardiamyocyte homeostasis that are expected to cause dysfunction and finally irreversible injury. Currently, there are no tools available that allow selective targeting of hemichannels, as all known pharmacological blockers inhibit both GJs and hemichannels [26, 27, 68]. Furthermore, connexin knockout (KO) technology abolishes hemichannels, as well as GJ channels making this approach inappropriate to determine the role of hemichannels in cardiovascular disease.
Connexins are tetraspan membrane proteins that have two extracellular loops and one intracellular loop (CL). Synthetic peptides like Gap26 and Gap27 that mimic a short stretch of amino acids (AAs) on the extracellular loops have been introduced more than 15 years ago to inhibit GJs . These peptides are thought to interact with yet undefined sequence on the extracellular loops of the connexin protein thereby preventing the docking of two hemichannels . Gap26 and Gap27 peptides also inhibit unapposed hemichannels ; because the extracellular loops of unapposed hemichannels are unoccupied and freely available for interactions with these peptides, hemichannel inhibition often occurs before inhibition of GJs. Here we report on a peptide, called Gap19 that is identical to a short sequence present on the intracellular (cytoplasmic) loop of Cx43. Peptides mimicking cytoplasmic loop (CL) sequences have been used in the past as control peptides that do not inhibit GJs . In agreement, we found that Gap19 did not reduce GJ coupling as measured with dual whole-cell voltage-clamp and dye transfer assays. Surprisingly, Gap19 strongly inhibited plasma membrane Cx43 hemichannels as exemplified by ATP release/dye uptake studies and unitary hemichannel current measurements. Surface plasmon resonance (SPR) experiments demonstrated that Gap19 interacts with the Cx43 C-terminus (CT) and hemichannel inhibition was counteracted by a peptide identical to the last 10 AAs of the CT, indicating that Gap19 inhibition of hemichannels is caused by preventing intramolecular interactions of the CT with the CL, which are essential for Cx43 hemichannel activities [49, 50]. Moreover, Gap19 inhibited the potentiation of unitary hemichannel currents in acutely isolated ventricular cardiomyocytes exposed to metabolic inhibition. In line with this finding, the peptide protected against myocardial cell swelling and cell death in in vitro cardiac ischemia/reperfusion studies and modestly limited the infarct size in in vivo cardiac ischemia/reperfusion in mice. Importantly, Gap19 had no effect on pannexin-1 (Panx1) or Cx40 hemichannels. Thus, Gap19 emerges as a novel tool to specifically block Cx43 hemichannels without inhibiting GJs, allowing in vitro and in vivo work aimed at determining the role of hemichannels in cardiac disease models, as well as in other tissues and organs that display a prominent Cx43 expression.
An expanded Methods section is provided in the Online Resource.
GJ coupling was investigated by dye coupling and electrophysiological studies. Dye coupling was assessed by fluorescence recovery after bleaching (FRAP) and scrape loading of dye transfer (SLDT) . For FRAP, confluent cell cultures were loaded with 5-carboxyfluorescein diacetate acetoxymethylester (CFDA-AM, 20 μM). Fluorescence within a single cell was photobleached by spot exposure to 488 nm Argon laser light. The fluorescent intensity in the bleached cell was followed over a 5-min period and quantified at the end of this period as the percentage of recovery relative to the starting level before bleaching. SLDT was performed by making a linear scratch across a confluent monolayer of cells in the presence of 6-carboxy fluorescein (6-CF, 0.4 mM). A fluorescence diffusion profile perpendicular to the scratch was recorded and the spatial constant of mono-exponential fluorescence decrease was determined as a measure of GJ coupling. The junctional electrical conductance between cells was measured as described in Ref. .
Hemichannel opening was investigated by ATP release studies and electrophysiological measurements of unitary hemichannel activity. ATP release was detected using a luciferin/luciferase assay kit (product no. FL-AA; Sigma-Aldrich, Bornem, Belgium) as previously described . Hemichannel unitary currents were measured as described in Ref. . For measurements in cardiomyocytes, KCl in the pipette solution was replaced by CsCl .
Cardiomyocytes for patch-clamp experiments were isolated from left ventricles of domestic pigs (14–17 weeks) by enzymatic digestion . The mouse cardiomyocytes used for in vitro ischemia/reperfusion studies were isolated by excising C57/BL6 mice (< 3 months) hearts followed by cardiomyocyte isolation as described in Ref. .
Primary cardiomyocytes were isolated from mice as described in Ref. . Cells were dispersed in isolation solution (0.025 mM Ca2+) by gentle agitation. Ca2+ in the solution was then gradually increased in small steps to 1 mM. Cardiomyocytes, sedimented to the bottom of a 10-ml tube, were exposed during 120 min to hypoxic (N2-gassed), glucose-deprived acidic (pH 6.5) solution [oxygen-glucose deprivation (OGD)/acidosis solution] on top of which a layer of mineral oil was added. Reperfusion consisted of removing the mineral oil and replacing the OGD/acidosis solution by a normoxic solution (pH 7.4) for 3 min. After normoxia or OGD/acidosis, a 10-μl cell sample was taken and resuspended for 5 min in control solution with 0.5 % trypan blue. Images of cell morphology were obtained at 100× magnification on a Leica DMLB microscope (Leica, Bensheim, Germany). Cell viability was quantified as the percentage of rod-shaped, unstained cells over the total cell population by an examiner blinded to the different conditions. A total of at least 1000 cells were counted per group.
For volume measurements, the cardiomyocytes were loaded with 2 μM calcein-AM in normoxic solution (15 min, 35 °C) and were then scanned with a confocal microscope (Zeiss axiovert 100 M, Jena, Germany) at 40× magnification and Z-stacks were taken every 2 μm in 15 cells per group. The cell volume was expressed relative to the cell volume under normoxic conditions. Exposure to OGD/acidosis was 60 min and the normoxia/reperfusion condition was applied 15 min in these experiments.
The experiments were approved by the regional ethical committee. C57/BL6 mice were subject to 30 min ischemia and 120 min reperfusion . The area at risk was determined by Evans Blue and the ischemic zone was visualized by 2,3,5-triphenyl tetrazolium chloride staining. Gap19 (MW 1161.44 Da) was intravenously administered at a concentration of 25 mg/kg, which corresponds to ~250 μM, assuming distribution in the blood volume that is approximately 8 % of the body weight (blood volume values were in the 1.6–2 ml range). The 250-μM concentration gives an expected 97 % inhibition based on the data presented in Fig. 1d.
The data are expressed as mean ± SEM, with ‘n’ denoting the number of independent experiments. In the in vitro and in vivo ischemia experiments, ‘n’ corresponds to the number of animals. Comparisons between two groups were done with a two-tailed unpaired t test; comparison of more than two groups was done with one-way ANOVA and a Bonferroni post hoc test; in the in vivo ischemia experiments a Fisher’s Least Significant Difference test was used. A p value < 0.05 was considered as indicating statistical significance. In the graphs, statistical significance is indicated with a single symbol (* or #) for p < 0.05, two symbols for p < 0.01 and three symbols in case of p < 0.001.
Gap19 is a synthetic nonapeptide corresponding to AAs 128–136 in the second half of the CL of Cx43 and is part of the so-called L2 region  (Fig. 1a and Online Resource Table S1). Furthermore, AAs 130–136 are part of a sequence (AAs 130–139) that is important for CL interactions with the CT tail of Cx43 [7, 22, 31]. Gap19 contains the KKFK sequence that is a known cell-membrane translocation motif that facilitates plasma membrane permeability . This may explain the higher uptake of Gap19 tagged with fluorescein isothiocyanate (FITC) in C6 glioma cells stably transfected with Cx43 (C6-Cx43) compared to fluorescein alone (Fig. 1b). We determined the effect of Gap19 on ATP release triggered by an elevation of the cytoplasmic Ca2+ concentration ([Ca2+]i) caused by exposing C6-Cx43 cells (plated at low density) to the Ca2+ ionophore A23187 (2.2 μM, 5 min, Fig. 1c, left bars), which, as reported, results in transient elevation of [Ca2+]i to ~500 nM [19, 49]. [Ca2+]i-triggered ATP release in C6-Cx43 cells depends on Cx43 hemichannels because Cx43 knock-down  or transfection of C6 cells with the empty vector (pLTR) do not display these responses (Fig. 1c, middle bar). Online Resource Fig. S1A shows Cx43 expression in C6-pLTR and C6-Cx43 cells. Gap19 inhibited [Ca2+]i-triggered ATP release (Fig. 1c, right bars) in a concentration-dependent manner with a half-maximal effect (IC50) of ~47 μM (Hill coefficient = 2) (Fig. 1d). Gap19 (400 μM applied via the pipette solution) did not inhibit the junctional conductance determined in Novikoff cell pairs (endogenously expressing Cx43) making use of dual-cell voltage-clamp (Fig. 1e). Cell-to-cell dye transfer studies with FRAP in C6-Cx43 cells confirmed that 30 min exposure to Gap19 (200 μM) had no effect on dye coupling (Fig. 1f, ,g).g). We verified higher Gap19 concentrations in FRAP experiments and found that 1 h exposure to concentrations of 10 μM to 1 mM Gap19 had no effect on dye coupling (Online Resource Fig. S1B). In fact, 24–48 h exposure to 200 μM Gap19 was found to promote dye coupling in FRAP studies (Fig. 1g). ATP release was still completely blocked following 24 h incubation with 200 μM Gap19 (inset to Fig. 1g). Gap19 did not inhibit ATP release triggered by exposure of C6 cells stably transfected with pannexin-1 (C6-Panx1) to high extracellular potassium concentration ([K+]e = 143 mM) (Fig. 1h). High [K+]e is a known stimulus for Panx1 hemichannel opening  and control experiments confirmed that the triggered ATP release was suppressed by low concentrations of carbenoxolone (10 μM, 30 min) and 10Panx1 peptide (200 μM, 30 min), two blockers of Panx1 hemichannels [42, 48], but not by scrambled 10Panx1 (Scr10Panx1) [70, 76] (Fig. 1h). Exposure of C6 wild type (WT) cells to high [K+]e did not trigger ATP release above baseline (Fig. 1h—Online Resource Fig. S1A illustrates Panx1 expression in the cells used). Furthermore, Gap19 did not influence [Ca2+]i-triggered ATP release in HeLa cells expressing Cx40 (Fig. 1i), a major Cx expressed in the atrium and cardiac conducting system .
Gap19 contains two AAs known to be mutated in certain types of oculodentodigital dysplasia (ODDD), a Cx43-linked genetic disease [63, 65]. I130T mutation is linked to neurological abnormalities and associated with decreased GJ dye transfer and non-functional hemichannels [39, 65]. Moreover, I130 is involved in the formation of hydrogen bonds  indicating that this AA might be important for Gap19 activity. K134E mutation decreases the single-channel conductance of GJ channels and interferes with normal GJ plaque formation [63, 65]. We found that Gap19 containing the I130A modification (Gap19I130A, concentrations as for Gap19) failed to inhibit [Ca2+]i-triggered ATP release (Fig. 2a). The cellular uptake of FITC-Gap19I130A was comparable to the uptake of FITC-Gap19 as illustrated in Fig. 1b. To determine the importance of neighboring AAs, we measured the effect of Gap19E131A and found it equally active as Gap19 (Fig. 2a). Since mutations of K134 may alter the membrane permeability of Gap19, we constructed a plasmid encoding Gap19 coupled via its N-terminus to a FLAG-tag (pFLAG-Gap19). In contrast to C6-Cx43 cells expressing pcDNA5/FRT-eGFP (empty vector), C6-Cx43 cells expressing pFLAG-Gap19 displayed no significant [Ca2+]i-triggered ATP release (Fig. 2b) and maintained normal GJ coupling measured with SLDT (Fig. 2c, ,d),d), similar to the results obtained with exogenously applied Gap19 peptide. In line with the results obtained with Gap19I130A (Fig. 2a), pFLAG-Gap19I130A failed to suppress ATP release (Fig. 2e). pFLAG-Gap19Q129A inhibited ATP release like pFLAG-Gap19, further supporting the notion that AAs neighboring I130 are not essential for hemichannel inhibition (Fig. 2e). Finally, pFLAG-Gap19K134A was ineffective as an inhibitor of ATP release (Fig. 2e). Thus, I130 and K134 are important AAs in the inhibitory effect of Gap19 on ATP release.
Since intramolecular interactions of the last ten AAs of the CT with the CL are important for Cx43 hemichannel function  and AAs 130–136 in the CL are involved in CT binding, we investigated whether Gap19 (AAs 128–136) blocks hemichannels by binding to the CT thereby preventing CT–CL interaction. Pre-incubating C6-Cx43 cells with membrane-permeable TAT-CT10 (100 μM, 30 min), a peptide corresponding to the last ten AAs of the Cx43 CT (Online Resource Table S1), followed by co-incubation with Gap19, completely abolished Gap19 inhibition of [Ca2+]i-triggered ATP release, while the peptide with reversed CT10 sequence (TAT-CT10reverse) did not (Fig. 2f). Mutant versions of TAT-CT10 were designed to determine the residues critical for neutralizing Gap19 activity. The choice for particular mutants was based on the data of Hirst-Jensen et al.  pointing to the importance of residues in the 376–379 domain for CT–CL interactions. Modifying D378 and D379 to alanine residues (TAT-CT10DD/AA) or P375 and P377 to glycine residues (TAT-CT10PP/GG) resulted in peptides that failed to abolish Gap19 activity, while modifying R374 and R376 to alanine residues (TAT-CT10RR/AA) had no effect (Fig. 2f). Similar results were obtained with pFLAG-Gap19 expression studies instead of exogenously applied Gap19 peptide (Online Resource Fig. S2). These results indicate that the two aspartate and two proline residues present in the last 10 AA-stretch of the CT are critical for Gap19-inhibitory actions.
It is known that the L2 region of the CL, in which sequence of Gap19 is located, is important for interactions with the CT [22, 31, 49]. We further explored Gap19-CT interactions with SPR and monitored the association of purified Cx43 CT tail (AAs 255–382) with biotin-Gap19 immobilized to a streptavidin-coated sensor chip. Immobilized biotin-L2 (AAs 119–144) and its reversed sequence version were used as positive and negative controls, respectively. These experiments displayed a clear association of the CT tail with Gap19 and L2. Interestingly, the binding of the CT tail to Gap19 seemed stronger as compared to L2 (Fig. 3a, ,b).b). The CT-Gap19 association signal increased at higher concentrations of the CT tail indicating a bona fide specific interaction. From these results, an estimated KD of ~2.5 μM was obtained (Fig. 3c, ,dd).
We used HeLa cells stably transfected with Cx43 (HeLa-Cx43) to determine whether Gap19 inhibits unitary currents through hemichannels in voltage-clamp experiments , performed on solitary non-coupled cells, in the presence of extracellular Ca2+ and Mg2+, and under conditions of K+-channel blockade (Fig. 4a). Application of voltage ramps showed that hemichannel currents appeared at potentials above +50 mV (Fig. 4b) and we chose +70 mV as a command voltage for subsequent experiments. Applying a depolarizing voltage step from −30 to +70 mV triggered unitary current events characterized by a conductance of ~220 pS (Fig. 4c, ,d),d), which corresponds to the single-channel conductance of Cx43 hemichannels . The activity of unitary events was reduced when Gap19 (400 μM) was present in the pipette solution, while Gap19I130A had no effect (Fig. 4c). Unitary event activity often displayed multiple superimposed stepwise channel openings, which were less frequent with Gap19. Figure 4d (graphs in left column) illustrates that Gap19 decreased the number of peaks in the conductance histograms and increased the frequency of the closed state. Figure 4d (graphs in right column) demonstrates that Gap19 did not influence the time constant of open dwell-time distributions while clearly decreasing the number of channel openings. Integration of the current versus time traces, which yields the membrane charge transfer (Qm) associated with unitary current activity, showed that Gap19 significantly suppressed Qm to ~25 % of the control level, whereas Gap19I130A had no effect (Fig. 4e). Online Resource Fig. S3A illustrates that the effect of Gap19 was rapid within 1 min. Gap19 inhibited Qm in a concentration-dependent manner with an IC50 of ~6.5 μM (Hill coefficient = 2) (Fig. 4f). By contrast, Gap19I130A had no effect unless applied at 1 mM concentration at which point the flat Qm curve displayed a sharp decrease, presumably as a result of steric block of the channel pore (Fig. 4g). A separate set of experiments was performed in cell-attached patch mode, to determine the effect of Gap19 on the open probability (Popen) of single hemichannels. The Popen was 0.01 ± 0.0004 for control recordings (Vm steps to +70 mV) and 0.0009 ± 0.0003 with Gap19 present in the bath (200 μM, 30 min pre-incubation) demonstrating a significant ~10-fold reduction by Gap19 (p < 0.0001, n = 5).
We sought to identify single-channel Cx43 hemichannel activity in ventricular cardiomyocytes acutely isolated from adult pig heart, under recording conditions as applied in the experiments on HeLa-Cx43 cells but with CsCl instead of KCl as the main pipette salt. Voltage steps from −70 mV to Vm in the range of +10 to +100 mV triggered single-channel current activity from Vm = +30 mV on, with a single-channel conductance of ~200 pS as judged from conductance histograms (Fig. 5a). When Gap19 was added to the recording pipette (100 μM), unitary current activity was strongly decreased and voltage steps to more positive Vm were necessary to observe unitary events (Fig. 5a). Figure 5b illustrates that, under control conditions without Gap19, stepping back from positive Vm (applied 30 s) to −70 mV caused clearly discernible single-channel activity that was present during 100–200 ms; all-point histogram analysis indicated a ~200 pS unitary conductance (Fig. 5b). The recordings of single-channel activity at positive (during the Vm step) and negative (tail currents) voltages allowed to construct a plot of unitary membrane current (Im) as a function of Vm (Fig. 5c). This open hemichannel Im − Vm plot demonstrated a linear relation characterized by a reversal potential of ~0 mV and a single-channel slope conductance of 196 ± 3.8 pS (n = 7) (Fig. 5c). We analyzed the voltage-dependence of unitary current activation under control and in the presence of Gap19 (100 μM) and found that the activation curve was shifted to the right by ~30 mV (Fig. 5d). Gap19 thus significantly increased the Vm threshold for hemichannel activation. Although the unitary activity was strongly depressed by Gap19, sufficient data (at strong positive Vm and in the tails at negative Vm) were available to construct an Im − Vm plot (Fig. 5e). This analysis revealed a reversal potential of ~0 mV and a single-channel slope conductance that was very similar to the ones obtained from control conditions (204 ± 10.4 pS; n = 6). Thus, Gap19 inhibits unitary currents by shifting the voltage-dependence of hemichannel opening without affecting the open channel properties. Hemichannel inhibition by Gap19 in cardiomyocytes was slower as compared to recordings obtained in HeLa-Cx43 and it took several minutes to attain maximal effect (Online Resource Fig. S3B). This is likely caused by the lower concentration of Gap19 (100 μM versus 400 μM for Hela Cx43), the larger cell size of cardiomyocytes and a possibly more restricted intracellular diffusion.
Metabolic inhibition (MI) with mitochondrial uncoupler carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) combined with the glycolysis inhibitor sodium iodoacetate (IAA) (‘chemical ischemia’) activates hemichannel-related macroscopic currents and dye uptake in various cell types endogenously expressing Cx43 [13, 34, 38, 52]. We applied FCCP/IAA (5 μM and 1 mM respectively) and found it to strongly augment the unitary current activity recorded with Vm steps to +40 mV (Fig. 5f). Most notably, when Gap19 was added in the pipette (100 μM), unitary activity was completely absent. Figure 5g summarizes average data from such experiments and shows that unitary current progressively increased upon prolonged exposure to MI while intracellular application of Gap 19 abolished single-channel events throughout the recording.
It has been reported that ischemic conditions trigger Cx43 hemichannel opening, possibly mediated by changes in the phosphorylation status of Cx43 and the generation of reactive oxygen and nitrogen species [28, 51, 54, 66]. We explored whether Gap19 could prevent possible deleterious effects of uncontrolled opening of hemichannels, thereby reducing ischemia/reperfusion damage to the myocardium in in vitro and in vivo studies. Exposing isolated cardiomyocytes to 120 min OGD and acidosis (pH 6.5) followed by a switch to control/normoxic solution (3 min) to simulate reperfusion resulted in extensive cell death (example image shown in Online Resource Fig. S4) with 10.9 % of the cardiomyocytes remaining viable (Fig. 6a). Cardiomyocytes exposed to normoxic solution over the same time period as OGD/acidosis showed no decline in survival (Fig. 6a). Pre-treatment of the cells with Gap19 (250 μM, 30 min) followed by OGD/acidosis (with Gap19 present) and reperfusion (normoxia) increased the viability to 14.6 % thus enhancing the viability by approximately one-third as compared to treatment with vehicle-only (Fig. 6a). Exposing isolated cardiomyocytes to OGD/acidosis (60 min) followed by reperfusion (15 min) resulted in significant cell swelling that was counteracted by Gap19 (treatment as in the viability assays) (Fig. 6b). Gap19 did not affect the cell volume under normoxic conditions (99.1 ± 1.8 % compared to 100 ± 2.2 % in control, n = 5). Gap19I130A exhibited some protective effect against cell death and cell swelling but the effect of Gap19 was significantly stronger (Fig. 6a, ,b).b). We performed myocardial ischemia/reperfusion experiments in mice in vivo, by applying a 30 min ligation of the left anterior descending (LAD) coronary artery followed by reperfusion (120 min). Gap19 was intravenously injected 10 min before ligation at a dose of 25 mg/kg, corresponding to an estimated 250 μM concentration when distributed in the blood volume. This significantly reduced the infarct size to 51.2 %, compared to 63.8 % infarct size in vehicle-treated control animals, i.e. a reduction by approximately one-fifth. Gap19I130A had no significant effect compared to control (Fig. 6c).
Our data show that Gap19 blocks Cx43 hemichannels without inhibiting GJs, at micromolar concentrations (IC50 of 6.5 μM) when applied intracellulary. This corresponds well to the ~2.5 μM KD value for interaction of Gap19 with its intracellular target which is the Cx43 CT tail. The IC50 of extracellulary applied Gap19 is higher, presumably because of incomplete permeation through the plasma membrane. Online Resource Fig. S5 shows that the IC50 of Gap19 linked to the TAT translocation motif, to further improve its membrane permeability, is ~7 μM. Thus, micromolar intracellular Gap19 concentrations interact with the CT and inhibit hemichannels. Hemichannel inhibition is not caused by steric block of the channel pore by the peptide, because (1) the mutant peptide Gap19I130A had no effect unless applied at 1 mM concentration, (2) CT peptide (TAT-CT10) removed Gap19 inhibition, (3) the single hemichannel conductance was not altered by Gap19, and (4) Cx43-based GJs were unaffected. The effect of Gap19 was selective as it had no effect on hemichannels composed of Cx40 or Panx1. Cx40 is a major Cx in the atria, with a long CT as Cx43 and a slightly lower MW; Panx1 hemichannels have also been reported to be expressed in atrial cardiomyocytes . The selectivity of Gap19 probably relates to the fact that the intracellular domains of the Cx protein are the least conserved region in contrast to the extracellular domains .
The molecular basis of the differential effect of Gap19 on hemichannels and GJs is related to the binding of this peptide to the CT tail which prevents CT–CL interaction. Disrupting CT–CL interaction results in reduced hemichannel openings and ATP release as demonstrated here, while avoiding closure of GJ channels as reported by others  and confirmed in the present experiments. The mechanisms underlying this differential regulation of hemichannels and GJs by CT–CL interactions are still unclear. However, it is important to notice that disrupting CT–CL interactions by CT-truncation of Cx43 has similar consequences: GJs remain functional , while hemichannels become resistant to activation [18, 49]. We anticipate that nonjunctional hemichannels (closed) may adopt different conformations as compared to those incorporated into GJs (open) [72, 73]. Interactions between subunits during docking of two hemichannels are indeed likely to result in conformational changes of the Cx protein and thereby alter gating [10, 23]. Another element that may contribute is the fact that hemichannels and GJ channels are differentially distributed over plasma membrane domains with slightly different properties, for example lipid rafts [3, 41]. Alternatively, Gap19 treatment may result in a reduction of the hemichannel population in the nonjunctional plasma membrane, e.g. due to accelerated incorporation into GJs or internalization. However, we found that the density of unapposed/nonjunctional hemichannels rather increased upon ectopic expression of pFLAG-Gap19 in C6-Cx43 cells (Online Resource Fig. S6). This figure also demonstrates increased phosphorylation of the hemichannel population with pFLAG-Gap19, which may well contribute to hemichannel inhibition54 and may counteract decreased phosphorylation associated with ischemia [34, 38, 66].
Limited data are currently available demonstrating that nonjunctional membrane hemichannels in cardiomyocytes can open under certain conditions. The observation of a [Ca2+]i-dependent non-selective, macroscopic membrane current in rabbit ventricular myocytes in response to MI initially suggested Cx43 hemichannel opening in the heart [34, 38]. More recent findings report a marked increase of cell permeability based on ATP release and dye uptake associated with activated hemichannels in simulated ischemia of neonatal cardiomyocytes [11, 33, 66]. Here we demonstrate in ventricular cardiomyocytes single-channel plasma membrane currents with a unitary conductance of ~200 pS in response to stepping Vm to +30 mV and above. The biophysical properties of the observed unitary activity point to Cx43 hemichannels: (1) the single-channel conductance determined from histogram analysis or Im − Vm plot measurements was similar as observed in Hela cells overexpressing Cx43 and corresponds to approximately twice the single-channel conductance of Cx43 GJ channels; (2) currents reverse around 0 mV, indicating a non-selective ion channel; (3) unitary currents are inhibited by Gap19 as in HeLa-Cx43 cells. Interestingly, unitary hemichannel activity was also present in the tail currents during repolarization from positive voltage steps which may be caused by increased [Ca2+]i that is known to potentiate hemichannel opening if < 500 nM  or slow deactivation kinetics. Gap19 inhibited unitary hemichannel opening events and this was mainly related to an increased (more positive) voltage threshold for hemichannel opening and not the consequence of channel pore block (no change in the single-channel conductance). Thus, Gap19 appears to alter the voltage-sensitivity of Cx43 hemichannel gating as a consequence of disrupted CT–CL interactions.
Of note, Gap19 inhibition of ATP release was complete (Fig. 1d) while inhibition of unitary hemichannel currents was incomplete at high concentrations (Fig. 4f). This may point to a more pronounced inhibitory effect on the passage of large MW substances as compared to ions. Such differences in the potency of inhibition have been reported by others: phosphorylation of Cx43 hemichannels by PKC limits the passage of sucrose (MW 342 Da, MW of ATP is 507 Da) but not of the smaller ethyleneglycol (MW 62) . An alternative explanation is that ATP release may behave in a non-linear manner (for example by Ca2+-or ATP-induced ATP release) as suggested by previous studies demonstrating complete inhibition of ATP release with only a 50 % reduction in Cx43 expression .
The unitary hemichannel currents were strongly potentiated by imposing MI. Earlier observations in a variety of primary cells endogenously expressing Cx43 have suggested elevated hemichannel activities during chemical ischemia [13, 52, 75]. These studies are, however, less conclusive as the evidence was based on hemichannel-permeable dye uptake studies (< 1 KDa) or macroscopic current measurements. Here we provide detailed analysis at the highest resolution in intact ventricular cardiomyocytes, demonstrating activation of hemichannel unitary currents under MI. Interestingly, our data demonstrate stimulated unitary Cx43 hemichannel activity at +30 mV and above, which is in the range of the +30–40 mV attained during the plateau phase of the cardiac action potential. Several molecular mechanisms have been put forward to underlie increased hemichannel activity: (1) ATP depletion in metabolically inhibited cells affects the phosphatase/kinase balance, leading to Cx43 dephosphorylation which favors an open channel conformation [2, 37]; (2) S-nitrosylation may contribute to enhanced channel activities [52, 53] and (3) an increased fraction of hemichannels at the cell surface as observed for both Cx32 and Cx43 hemichannels within minutes after MI could also account for stimulated unitary current activities [53, 59]. Most notably, the unitary hemichannel currents activated by depolarization to +40 mV and promoted by MI were completely blocked by Gap19. This complete block is likely the result of a Gap19-induced increase in activation potential for hemichannel opening, which is shifted ~30 mV in the positive direction and therefore results in complete disappearance of current activity at +40 mV. Thus, ischemic conditions and depolarization activate hemichannels and this is inhibited by Gap19. The in vitro and in vivo cardiac ischemia/reperfusion experiments demonstrated that Gap19 significantly counteracted cell swelling, cell death and development of myocardial infarction. Taking together the facts that excessive opening of hemichannels may accelerate cell death [14, 58] and that Gap19 counteracts hemichannel opening in cardiomyocytes suggests that the improved outcome after ischemia/reperfusion is related to inhibition of Cx43 hemichannels present in the sarcolemma.
Not only Cx43 proteins embedded in the sarcolemma, but also those reported in mitochondria have been proposed to contribute to the cardioprotective effect of ischemic preconditioning [30, 62], a procedure of repeated exposure to sub-lethal ischemic conditions [46, 57, 61, 80]. Cx43 has been reported to be located in the inner mitochondrial membrane [5, 6, 44, 56], possibly in the hemichannel configuration, and appears to influence mitochondrial K+ fluxes thereby providing cardioprotective effects . It is important to realize that the putative protective role of mitochondrial Cx43 has only been demonstrated in the context of ischemic preconditioning ; currently, there is no evidence that this could play a role in a single episode of ischemia/reperfusion as applied in the present study.
Considering all evidence, the present data strongly point to the inhibition of plasma membrane hemichannel opening by Gap19 as the mechanism responsible for its protective effects against cardiac ischemia/reperfusion injury. The degree of protection, however, appears to be modest as opposed to recent evidence indicating a stronger effect of two extracellular loop mimetic peptides, Gap26 and Gap27 [28, 29]. Gap26/27 peptides target Cx43 channels but may also influence those composed of Cx37 and Cx40 [16, 25, 79]. The lesser degree of specificity toward different connexins is related to the fact that the extracellular loop sequence they mimic is highly conserved among different connexins. Thus, Gap26/27 peptides may have targets other than Cx43 hemichannels that confer additional protective potential. Most notably in this context is the fact that Gap19 does not inhibit GJs and thereby circumvents potential pro-arrhythmogenic effects of decreased GJ coupling during ischemia/reperfusion [32, 60]. Obviously, it would be interesting to further substantiate the effects of Gap19 on hemichannels in the in vivo situation; however, single-channel measurements under those conditions are extremely difficult to perform because the large degree of cardiomyocyte coupling via GJs precludes reliable space clamp conditions.
The distinctive effects of Gap19 on GJs and hemichannels are of fundamental importance: GJs and hemichannels are composed of the same Cx subunits and KO animal technology influences both channel types equally. Here we report Gap19 as a novel tool allowing selective studies on the role of Cx43 hemichannels in normal and diseased heart. In addition, blocking of hemichannels with Gap19 opens an avenue for therapeutic applications, limiting cellular injuries during ischemia/reperfusion while preserving electrical and metabolic cell–cell communication that are vital for the normal function of the myocardium.
Special thanks to K. Leurs, K. Vermeulen and K. Welkenhuyzen for superb technical support. We express our gratitude to Dr. B. Himpens for support, to Dr. P. Zimmermann for the use of the Biacore 2000, to Dr. P. Sorgen for providing the pGEX6p2-Cx43CT plasmid, to Dr. G. Antoons for support with experiments on pig myocytes, and to Dr. D. Laird and Dr. S. Penuela for providing the anti-Panx1 antibody. We are very grateful to A. Gadicherla and Dr. B. Nilius for critically reading and commenting the manuscript. Research supported by the Fund for Scientific Research Flanders, Belgium (FWO, grant nos. G.0354.07, G.0140.08, 3G.0134.09 and G.0298.11 N to L.L. and G.0545.08 to G.B.), the Interuniversity Attraction Poles Program (Belgian Science Policy, project P6/31 and P7/10 to K.R.S and L.L., and P7 to J.T. and G.B.), the Concerted Actions program at KULeuven (grant no. GOA/09/012 to G.B.), the German Research Foundation (Schu 843/7-2 to R.S.), the BFH grant (grant no. PG/01/1298 to W.H.E), the Heart & Stroke Foundation of BC & Yukon and the Canadian Institutes of Health Research to C.C.N. and NIH grants (HL084464 and NS072238 to F.F.B.).
G. Bultynck, L. Leybaert share senior authorship.
T. Miura, Sapporo, Japan served as guest editor for the manuscript and was responsible for all editorial decisions, including the selection of reviewers. The policy applies to all manuscripts with authors from the editor’s institution.
Conflict of interest None.