Anti Gβ antibody was obtained from Santa Cruz Biotechnology. Anti CX43 antibody and IGEPAL CA-630 was obtained from Sigma. Gallein was purchased from Calbiochem. Cleaved caspase-3 andtibody was obtained from Cell Signaling Technology. Recombinant Gβ₁γ₂ was obtained from Calbiochem-Merck Millipore. MTT (3-(4,5-methylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) was purchased from Dojindo Molecular Laboratories (Kumamoto, Japan).

The peptides were synthesized and purified by Invitrogen and peptide identity was verified by matrix-assisted laser desorption ionization mass spectrometry. The peptides were dissolved in aliquots (10 mM) and immediately frozen at −70°C.

The segments of AGS8 (DQ256268) were amplified by PCR and fused in frame to GST in the pGEX-6T vector (Amersham Biotech). Each GST-partial-AGS8 fusion proteins were expressed in bacteria (Escherichia coli BL21, Amersham Biotech) and purified on a glutathione affinity matrix. The GST fusion protein was eluted from the resin, and glutathione was removed by desalting to allow a solution-phase interaction assay [19]. Protein interaction assays and immunoblotting were performed as described previously [19], [20].

Cardiomyocytes were prepared from the hearts of 1-3-day-old Wistar rats as described previously [13]. The neonates were deeply anesthetized with pentobarbital sodium (100 mg/kg) and decapitated for cardiac tissue harvesting. The ventricular cardiomyocytes were then enzymentically dissociated and seeded at 1.0×10⁵ cell in 24 mm or 4.0×10⁵ cell in 35 mm plates. Prepared cardiomyocytes were cultured in DMEM/F12 including Insulin–Transferrin–Selenite (ITS), 100 units/ml penicillin, 100 mg/ml streptomycin, and 10 mM glutamine, in 5% CO₂ at 37°C. In some experiments, cardiomyocytes were incubated in a hypoxic incubator (MODEL9200, Wakenyaku, Kyoto, Japan) equilibrated to 1% O₂, 5% CO₂, and 94% N₂ at 37°C. The cells were subjected to two different hypoxic challenges. In the first protocol, cardiomyocytes were exposed 3 times to 30 min of hypoxia with intermittent 30-min periods of normoxia to capture the early events caused by hypoxia in the living cells before they died. Particularly, internalization of cell-surface connexin 43 and changes in the permeability of connexin were analyzed. In the second protocol, cardiomyocytes were exposed to 1% O₂ for 6 h followed by 12 h of normoxia to induce hypoxia-mediated apoptosis. At the end of the second protocol, apoptotic cell death was analyzed. In some experiments, approximately 24 h after preparation, peptides were delivered to cardiomyocytes by using PLUSin (Polyplus, NY, USA) according to the manufacturer’s instructions. Briefly, ∼1.0 µg peptide was incubated with 2.0 µl of PLUSin in 100 µl supplied buffer for 15 min and then added the mixture to each dishes. This preparation typically provided 0.5 to 1 µM of peptide in the culture medium. The amount of peptide and the incubation time for peptide-delivery were optimized by analyzing incorporation of fluorescein isothiocyanate (FITC)-conjugated peptide into cardiomyocytes. In the condition used in this study, the chemical reagent for peptide delivery did not influence the number of living cells analyzed by trypan blue stain within 4 h treatment (0.75 µM peptide; 102.0±6.7%, 1.5 µM peptide; 90.5±5.0% versus no treatment control, not statistically significant, n = 4) [21].

Immunostaing of cultured cardiomyocytes was performed as described previously [13]. Briefly, cells were seeded on 24×24 mm polylysine-coated coverslips. Cells were fixed with 4% paraformaldehyde for 15 min and then incubated with 0.2% Triton X-100 for 5 min. After 1 h incubation of 5% normal donkey serum, cells were incubated with primary antibodies for 18 h at 4°C. Following 1 h incubation of secondary antibody (goat anti-mouse AlexaFluor 488 or goat anti-rabbit AlexaFluor 594, highly cross-absorbed, Molecular Probes), cells were incubated with 1 µg/ml 4′,6′-diamidino-2-phenylindole, dihydrochloride (DAPI) (Molecular Probes) in PBS for 5 min. Slides were then mounted with glass coverslips with ProLong Gold antifade reagent (Invitrogen). Images were analyzed by deconvolution microscopy (TE2000-E, Nikon, Tokyo, Japan). Obtained images were deconvoluted using NIS-Elements 3.0 software (Nikon) with a “no neighbors” deconvolution algorithm. All images were obtained from approximately the middle plane of the cells.

Up take of CX43 permeable fluorescence dye Lucifer Yellow (LY) (Molecular Probes) was performed as described previously [13]. Briefly, cardiomyocytes were incubated with 1 mM Lucifer Yellow (LY) (Molecular Probes) for 30 min. The fluorescence of LY was determined by fluorescence microscopy (B-3A filter, TE2000-E, NIKON, Tokyo, Japan) following removal of incorporated LY and rinse with PBS. The signal intensity was quantified in 10 randomly selected fields (10×20) using NIS-Elements 3.0 software (NIKON, Tokyo, Japan). The non-specific binding of LY was determined in the presence of a connexin hemichannel blocker, 50 mM of Lanthanum (Sigma-Aldrich).

In situ labeling of fragmented DNA in cardiac myocytes was detected by TACS2 TdT-Blue Label In Situ Apoptosis Detection Kit (Travigen, Inc., Gaitherburg, MD), that detects DNA breaks in genomic DNA by enzymatic incorporation of biotinylated nucleotides followed by the binding of streptavidin-peroxidase conjugates, according to the manufacture’s instructions. Briefly, myocytes were fixed with 3.7% formaldehyde in phosphate buffered saline (PBS) for 10 min and with 70% ethanol for 5 min and then incubated in proteinase K (0.02 mg/ml) at room temperature for 5 min. The cells were incubated with 2% hydrogen peroxide for 5 min and washed with labeling buffer consisting of 50 mM Tris (pH 7.5), 5 mM MgCl₂, 60 mM 2-mercaptoethanesulfonic acid, and 0.05% BSA, followed by 60 min of incubation at 37°C in labeling buffer containing 150 mM dATP, 150 mM dGTP, 150 mM dTTP, 5 mM biotinylated dCTP, and 40 U/ml of the Klenow fragment of DNA polymerase I. Untreated myocytes incubated with or without 2 mg/ml DNAse in the labeling buffer were used as positive or negative controls, respectively. The incorporated biotinylated dCTP was then detected with strepavidin-peroxidase conjugate and revealed in 0.5 mg/ml diaminobenzidine for 10 min. Nuclear brown staining was viewed under a light microscope.

Cardiomyocytes in 24-well culture plates were incubated in PBS (pH 7.4) containing 0.5 mg/ml MTT at 37°C. After for 2 h incubation, the PBS was removed and 100 µl of dimethyl sulfoxide (DMSO) was added to each well to solubilize dark blue formazan products. Absorbance of the colored solution was determined at 595 nm [22].

Immunoblotting and data analysis were performed as described previously [12], [13]. The luminescence images captured with an image analyzer (LAS-3000, Fujifilm, Tokyo, Japan) were quantified using Image Gauge 3.4 (Fujifilm). Data are expressed as mean ± S.E.M. from independent experiments as described in the figure legends. Statistical analyses were performed using the unpaired t test, F-test, one-way ANOVA followed by Tukey’s multiple comparison post-hoc test. All statistical analyses were performed with Prism 4 (GraphPad Software, USA).

Animal study was approved by the Institutional Animal Care and Use Committees of Yokohama City University.

To determine the region of interaction between Gβγ and AGS8, we divided the rat AGS8 (DQ256268) sequence into 6 segments and synthesized the segments as glutathione S-transferase (GST)-fusion proteins (Fig. 1A). Each of GST-AGS8 peptides was subjected to a pull-down assay to examine its interaction with purified Gβγ (Fig. 1B). The C-terminus of AGS8 (AGS8-C) successfully pulled down Gβγ, as documented in the previous manuscript [12]. Another segment, AGS8-254, also pulled down Gβγ to a lesser extent than AGS8-C, suggesting that a potential second Gβγ-interacting domain existed in this region. However, the remaining segments failed to pull down Gβγ.

Next, the bioactivity of each of the AGS8 regions was investigated by evaluating the activation the G-protein signaling pathway in Saccharomyces cerevisiae [12], [23]. This yeast strain lacked the pheromone receptor, but expressed mammalian Gαs in place of the yeast Gα subunit and provided a read-out of growth upon activation of the G-protein–regulated pheromone signaling pathway [23]. The peptide corresponding to each AGS8 domain was subcloned in a galactose-inducible vector and introduced into the yeast strain [24]. The bioactivity in the G-protein pathway was examined be evaluating the galactose-dependent growth of the transformed yeast. Although each segment of AGS8 was expressed in the yeast cells, AGS8-C was the only segment able to activate G-protein signaling (Fig. 1C). Thus, we focused on AGS-C and explored the Gβγ-interaction site in this domain.

The sequence of AGS8-C (A¹³⁵⁹ to W¹⁷³⁰ of rat ABB82299) was further divided into smaller fragments that were synthesized as GST-fusion proteins in bacteria. Two fusion proteins of GST-AGS8-C, namely, AGS8-C1 (A¹³⁵⁹– H¹⁴⁹³) and AGS8-C2 (A¹⁴⁹⁴ – T¹⁵⁸⁵), were successfully synthesized and migrated as expected on SDS-PAGE. While AGS8-C1 did not pull down purified Gβγ, AGS8-C2, which represented the FN3 domain, did pull down Gβγ, indicating the importance of the FN3 domain for the interaction of AGS8 with Gβγ (Fig. 1D).

To further determine the region of interaction between AGS8 and Gβγ, we prepared multiple 29- to 30-amino–acid peptides, the sequences of which were based on the amino acids from A¹⁴⁹⁴ to W¹⁷³⁰ of the rat AGS8 (ABB82299) (Fig. 2A). In a screen of 11 peptides by the GST-pull–down assay, which covered the entire region from A¹⁴⁹⁴ to W¹⁷³⁰ of the rat AGS8, we found two peptides, CP1 (A¹⁴⁹⁴PRNITVVAMEGCHSFVIVDWNKAIPGDV¹⁵²²) and CP9 (S¹⁵⁰⁸FVIVDWNKAIPGDVVTGYLVYSASYEDFI¹⁵³⁷) that effectively blocked the interaction of AGS8 with Gβγ in a dose-dependent manner (Fig. 2B). Quantitative analysis of immunoblots confirmed the dependence of the interaction of AGS8 with Gβγ on dose (Fig. 2C) and indicated that both peptides had similar IC50s (CP1∶6.73×10⁻⁶ M; CP9∶2.77×10⁻⁶ M). We first focused on CP9 and used this peptide as the AGS8-peptide in the following experiments.

The present data suggested that the first 45 amino acids of the FN3 domain represented by CP1 and CP2 were critical for AGS8 to interact with Gβγ and mediate signal to downstream molecules. The importance of this region for AGS8-mediated signaling was further examined in the yeast cells in which growth was linked to G-protein activation [23]. A deletion mutant of AGS8-C, which lacked the first 45 amino acids of this domain, failed to activate G-protein signaling, indicating the importance of this region for mediating AGS8-Gβγ signaling in the cell (Fig. 3).

In a previous study, we demonstrated that AGS8 was required for hypoxia-induced apoptosis of cardiomyocytes, which was associated with changes in the permeability of CX43 [13]. AGS8-Gβγ accelerates the internalization and degradation of channel protein CX43 under hypoxia, which results in decreased membrane permeability in the cardiomyocytes [13]. The change in localization and permeability of CX43 in the membrane is associated with hypoxia-induced apoptosis of the cardiomyocytes [14]–[16]. Therefore, we transferred the AGS8-peptide to cardiomyocytes and examined its effect on the internalization of CX43 induced by repetitive hypoxia (Fig. 4A).

The peptide was delivered to the cardiomyocytes by chemical reagent as described in “experimental procedures”. Treatment of the cells with the chemical reagent and FITC-conjugated AGS8-peptide showed that the peptide was successfully delivered into the cardiomyocytes (Fig. 4B). The effect of the AGS8-peptide on internalization of CX43 was determined in the cardiomyocytes after hypoxic stress. CX43 was observed on the surface of the cardiomyocytes under normoxia (Fig. 4B), and its presence was decreased after exposure of the cardiomyocytes to repetitive hypoxia in the untransfected control cells and the cells exposed to transfection reagent alone (Fig. 4B, 4C). However, the AGS8-peptide dramatically blocked the internalization and degradation of CX43 induced by repetitive hypoxia (Fig. 4B, 4C). When the effect of FITC-conjugated AGS8-peptide was examined, FITC was observed in the cardiomyocytes under normoxia as well as hypoxia, and the hypoxia-induced internalization of CX43 was inhibited in FITC-positive cells (Fig. 4B).

Loss of CX43 from the cell surface decreases influx and efflux of small molecules passing through CX43, and this effect is associated with apoptosis of the cardiomyocytes under hypoxia [13]–[16]. We next examined the ability of AGS8-peptide to block the change in permeability of CX43 by analyzing by the flux of the fluorescent dye, Lucifer Yellow (LY), which passes through CX43 as previously demonstrated [13]. LY in the culture medium was incorporated into cardiomyocytes under normoxia. The flux of dye was decreased after exposure of the cells to repetitive hypoxia in the control group as well as in the group exposed to transfection reagent alone (31.1±4.4%, 26.9±13.0%, respectively) (Fig. 5). However, the AGS8-peptide blocked the hypoxia-induced decrease in permeability in a dose-dependent manner. This observation is consistent with the immunofluorescence studies, in which CX43 was observed to remain at the cell surface after repetitive hypoxia in the presence of AGS8-peptide (Fig. 4).

To examine the effect of AGS8 peptide on apoptosis of the cardiomyocytes, cultured cardiomyocytes were sequentially exposed to 1% oxygen for 6 h, then to 12 h of normoxia to induce hypoxia-mediated cell death (Fig. 6A) [13]. Hypoxia/reoxygenation markedly increased the number of apoptotic cardiomyocytes, as determined by TUNEL or immunostaining of cleaved caspase-3, in the untransfected control cells and those exposed to transfection reagent alone (Fig. 6B and 6C). However, AGS8-peptide successfully inhibited hypoxia-induced apoptosis, indicating a protective effect in cardiomyocytes. Additionally, the data indicated that, under normoxia, the AGS8-peptide did not influence apoptosis or the permeability of CX43 (Fig. 6B and 6C).

The data presented thus far indicated that the AGS8-peptide blocked the AGS8-Gβγ interaction and the signaling events the downstream of AGS8-Gβγ. We next asked whether inhibition of the Gβγ-mediated signal generally had a cardioprotective effect as was observed with the AGS8-peptide. Gallein is an inhibitor of Gβγ-mediated signaling that occupies a “common” interaction surface of Gβγ and inhibits the interaction of Gβγ-regulated proteins with Gβγ [25]. Thus, gallein is expected to block a wide range of Gβγ signals in cells, including CX43 regulation mediated by AGS-Gβγ. We previously demonstrated that gallein completely inhibits hypoxia-induced internalization of CX43 at 100 µM, as observed in the knockdown of AGS8, but not at 1 µM [13].

Here, we first examined the influence of the AGS8-peptide and gallein on the viability of cells. Cardiomyocytes were incubated with gallein for 24 h, and their viability was determined by MTT assay. Even the lowest concentration of gallein (1 µM) caused damage in the cardiomyocytes, indicating that broad inhibition of Gβγ did not have a protective effect (Fig. 7). In contrast, the AGS8-pepetide did not show cytotoxicity at 1 µM, the concentration that completely blocked the AGS8-Gβγ–mediated signal event. This suggests that the AGS8-peptide is a promising candidate for protection of cardiomyocytes from hypoxia-mediated apoptosis.