C57BL6 mice were maintained in the animal facility of the Collège de France (Paris). Animal experimentation was carried out in accordance with the European Community Council Directives (2010/63/UE) and care was taken to minimize their suffering. The study with C57BL6 mice was approved by the bioethics committee of the Collège de France (Paris). Albino Wistar rats were obtained from the animal house of the University of Salamanca (Spain) and were used according to local and EU Ethical Committee guidelines. The study with Albino Wistar rats was approved by the bioethics committee of the University of Salamanca and Junta de Castilla y León (Spain).

Synthetic peptides (>85% pure) were obtained from GenScript (Piscataway, NJ, USA). YGRKKRRQRRR was used as the TAT sequence, which is responsible for the cell penetration of the peptides. The sequence of TAT-Cx43_266–283 was TAT-AYFNGCSSPTAPLSPMSP (Gangoso et al., 2014).

Adult mice were subjected to intracerebral injection as previously described (Koulakoff et al., 2008; Gangoso et al., 2012). Briefly, mice were deeply anesthetized by intraperitoneal injection of 0.3 mL of 2% avertin and were then subjected to a low-pressure injection of 1 nmol KA (Sigma-Aldrich, St. Louis, MO, USA) in 1 μL phosphate-buffered saline (PBS) in controls, 1 nmol KA plus 1 nmol TAT in 1 μl PBS, or 1 nmol KA plus 1 nmol TAT-Cx43_266–283 in 1 μl PBS under stereotaxic guidance (coordinates: −2.1 mm anteroposterior, 1.5 mm mediolateral and 0.6 mm dorsoventral from Bregma) aiming the injection into the right cerebral cortex. Animals were sacrificed by cervical dislocation at 7 days post-injection.

Cryostat sections (20 μm) prepared as previously described (Mei et al., 2010) were fixed for 30 min at room temperature with 4% paraformaldehyde, washed in PBS and pre-incubated for 1 h in PBS containing 0.2% gelatine and 0.2% Triton-X100. Brain sections were incubated overnight at 4°C with rabbit polyclonal antibody anti-GFAP (1:1000, G9269; Sigma) and mouse monoclonal anti-NeuN (1:500, MAB377; Merck Millipore, Madrid, Spain). After three washes with PBS, sections were incubated for 2 h at room temperature with their corresponding Alexa Fluor secondary antibodies (1:2000; Life Technologies, Carlsbad, CA, USA). After several washes, slices were mounted in Fluoromount (Southern Biotechnologies, Birmingham, AL, USA) and examined with an epifluorescence microscope (Eclipse E800; Nikon, Tokyo, Japan).

For quantification, the whole area of the lesion was analyzed using Fiji (Schindelin et al., 2012). Every 20 μm across the entire lesion (1500 μm), the damaged area was drawn in each slice with the freehand selection tool (Figure 1A). For drawing, we followed as criteria a reduction in NeuN staining within neurons compared with the contralateral region. The necrotic tissue was included as part of the lesion (Figure 1B). Sometimes the necrotic tissue was lost but that area was counted, assuming it was damaged tissue. Although some slices where lost during the sectioning process, all the conditions analyzed comprised at least 80% of the lesion. The areas across the whole lesion were plotted against their location using Prism software. Cavalieri’s principle was used to estimate the volume of the lesion (Jelsing et al., 2005). Briefly, the volume was estimated multiplying the summatory of the areas, obtained as described previously, by the distance between the sections analyzed along the rostro-caudal axis.

Primary cultures were obtained as previously described (Tabernero et al., 1993). For neuron cultures, rat fetuses at 17.5 days of gestation were delivered by rapid hysterectomy after cervical dislocation of the mother. Postnatal 1-day newborn rats were used to prepare astrocytes in culture. Briefly, animals were decapitated and their brains immediately excised. After removing the meninges and blood vessels, the forebrains were placed in Earle’s balanced solution (EBS) containing 20 μg/ml DNase and 0.3% (w/v) BSA. The tissue was minced, washed, centrifuged and incubated in 0.025% trypsin (type III) and 60 μg/ml DNase I for 15 min at 37°C. Trypsinization was terminated by the addition of Dulbecco’s Modified Eagle’s Medium (DMEM) containing 10% fetal bovine serum (FCS; Gibco, Life Technologies). The tissue was then dissociated by gently passing it eight times through a siliconized Pasteur pipette, and the supernatant cell suspension was recovered. This operation was repeated and the resulting cell suspension was centrifuged. The cells were then resuspended in DMEM containing 10% FCS and plated on Petri dishes coated with 10 μg/ml poly-L-lysine at a density of 10⁵ cells/cm². Cells were maintained at 37°C and 5% CO₂. One day after plating, 10 μM cytosine arabinoside was added to avoid glial cell proliferation on neuronal cultures.

For neuron-astrocyte co-cultures, the cell suspension obtained for neuron culture was plated at a density of 3.75 × 10⁴ cells/cm² on confluent 21 days in vitro (DIV) astrocytes. These co-cultures were maintained at 37°C and 5% CO₂ in DMEM + 10% FCS for 7 days and then different treatments were applied for 8 h.

For non-contact neuron-astrocyte co-cultures, the cell suspension obtained for neuron culture was plated at a density of 10⁵ cells/cm² in 12-well plates coated with 10 μg/ml poly-L-lysine. Cells were maintained at 37°C and 5% CO₂ and 1 day after plating, cytosine arabinoside was added to avoid glial cell proliferation. Eighteen DIV astrocytes were plated in 500 μL DMEM + 10% FCS on inserts containing polyethylene terephthalate filters with 1-μm pores (Merck Millipore) at 10⁵ cells/cm², whereas 1 ml DMEM + 10% FCS was added to the lower well. After 3 days, the medium of the astrocytes was changed and the inserts were placed on top of 4 DIV neurons with 25% of the medium changed. These non-contact co-cultures were maintained at 37°C and 5% CO₂ in DMEM + 10% FCS in the presence of the different treatments for 3 days.

All treatments were added to the culture medium and maintained at 37°C for the indicated times. The treatments were as follows: 50 μM TAT, 50 μM TAT-Cx43_266–283, 100 μM KA, 10 μg/ml lipopolysaccharide (LPS; Sigma), 200 μM carbenoxolone (CBX; hemichannel inhibitor, Sigma), 1 μM dasatinib (c-Src inhibitor; Selleck Chemicals, Munich, Germany), dimethyl sulfoxide (vehicle for dasatinib; 1 μl/ml), 20 μM 3-(2-carboxypiperazin-4-yl)propyl-1-phosphonic acid (CPP; NMDA receptor blocker), 200 μM Adenosine 5′-triphosphate, periodate oxidized sodium salt (oATP; P2X receptor blocker, Sigma) and 100 μM Brilliant Blue G (BBG; P2X7 receptor blocker, Sigma).

Cells were fixed with 4% (w/v) paraformaldehyde in PBS for 20 min and blocked for 30 min in antibody diluting solution (PBS containing 10% FCS, 0.1 M lysine and 0.02% sodium azide). Cells were then incubated overnight at 4°C with mouse anti-NeuN (1:100) and for 2 h with the secondary antibody anti-mouse labeled with Alexa Fluor 488 (A11029; Life Technologies) all prepared in antibody diluting solution containing 0.1% Triton-X100. Nuclei were stained with 4′,6′-diamidino-2-phenylindole (DAPI; 1.25 μg/ml; Invitrogen) for 10 min. Cells were then mounted using the Slowfade Gold Antifade Kit (ThermoFisher) and analyzed on a Nikon inverted fluorescence microscope connected to a digital video camera (DC100; Leica, Wetzlar, Germany). Negative controls carried out by omission of the primary antibodies resulted in absence of staining in all cases. At least six photomicrographs were taken from each plate. The number of nuclei (DAPI staining) and NeuN-positive cells were counted with ImageJ (NIH, Bethesda, MD, USA) on 8-bit images. The percentage of NeuN-positive cells was calculated from the total number of cells (DAPI staining).

Cells cultured at 37°C were incubated in the dark for 75 min with culture medium containing 0.5 mg/ml MTT (Sigma). The medium was then removed, and the cells were incubated for 10 min in the dark with dimethyl sulfoxide with mild shaking. Finally, the absorbance was measured at a wavelength of 570 nm using a microplate reader (Appliskan 2001; Thermo Electron Corporation, Thermo Scientific, Madrid, Spain).

Cultured cells were incubated with 5 μM ethidium bromide (EtBr) in HEPES-buffered salt solution (140 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose and 5 mM HEPES; pH 7.4) for 10 min. Cells were then washed with the same buffer, fixed with 4% paraformaldehyde in PBS and analyzed on a Nikon inverted fluorescence microscope connected to a digital video camera (Leica DC100). The EtBr fluorescence intensity within the nucleus was measured with the ImageJ software and the average of data from six images from different fields in the same culture was calculated to obtain the final measurement of dye uptake in each culture.

Gap junction permeability was determined by the scrape-loading/dye transfer technique as previously described (Herrero-González et al., 2010). Scrape-loading was performed by scraping the cell layer with a broken razor blade in a Ca²⁺-free ionic solution containing Lucifer yellow (LY; 1 mg/ml). LY is a low molecular weight (457 Da) fluorescent dye that can pass through the gap junctions of loaded cells to their neighbors. After 2 min, the dye solution was removed and the cells were carefully washed. Subsequently, 8 min after scraping, fluorescence photomicrographs were captured with a digital video camera (Leica DC100) connected to an inverted fluorescent microscope equipped with the appropriate filters (Diaphot, Nikon). At least six photomicrographs of the center of the dish were taken and the number of fluorescent cells per field were counted.

Results are expressed as the mean ± SEM of at least three independent experiments. Statistical analyses were carried out in Sigma Plot 11 (Systat Software) using an analysis of variance (one-way ANOVA), followed by the Holm-Sidak method for multiple comparisons at a significant level of p < 0.05.

Following the method previously described (Koulakoff et al., 2008), KA was cortically injected and after 7 days, brains were dissected, fixed and 20-μm frontal sections were cut along 1500 μm to cover the entire lesion (Figure 1A). Neuronal death was estimated by measuring the area in which neurons contain a reduced NeuN intensity when compared to contralateral region (Figure 1B), as described in “Materials and Methods” section. In agreement with previous results (Koulakoff et al., 2008), KA promoted neuronal death and reactive gliosis, characterized by the increase in GFAP expression (Figure 1B). Figure 1C shows neuronal death areas along the rostrocaudal axis found 7 days after KA injection.

Next, we analyzed the effect of TAT-Cx43_266–283 on neuronal death promoted by KA using TAT penetrating peptide as a control. TAT-Cx43_266–283 was injected with KA and its effect was compared with KA or KA+TAT. Our results showed that TAT-Cx43_266–283 diminished reactive gliosis as judged by GFAP staining (Figure 1D). More importantly, TAT-Cx43_266–283 prevented the reduction in NeuN-positive cells promoted by KA (Figure 1E).

Indeed, when the neuronal death area was estimated along the rostrocaudal axis (Figure 2), there was a strong reduction in neuronal death in animals injected with KA+TAT-Cx43_266–283 (Figure 2C1) compared to KA (Figure 2A1) or KA+TAT (Figure 2B1). Quantification of neuronal death areas along the rostrocaudal axis showed a reduction in the height and length of the curves in mice treated with KA+TAT-Cx43_266–283 (Figure 2C2) compared to KA (Figure 2A2) or KA+TAT (Figure 2B2). The area under these curves was calculated, using Cavalieri’s principle (described in “Materials and Methods” section) as an estimation of the volume of these lesions (Figure 2D). These results confirmed that TAT-Cx43_266–283 strongly reduced the extent of neuronal death promoted by KA injection (Figure 2D).

In order to gain insight into the neuroprotective mechanism of TAT-Cx43_266–283 we used cell cultures to address the contribution of each type of cell to this process. First, the effect of KA and TAT-Cx43_266–283 in neurons from primary cultures was analyzed. Surprisingly, TAT-Cx43_266–283 did not protect against neuronal death promoted by KA in these cultures (Figure 3), suggesting the participation of a glial cells partnership in the neuroprotective effect of TAT-Cx43_266–283. Indeed, when neurons were grown on top of astrocyte cultures, TAT-Cx43_266–283 prevented neuronal death promoted by KA (Figure 4). The number of NeuN-positive cells decreased in neuron-astrocyte co-cultures in the presence of KA (Figure 4A vs. Figure 4B). However, when KA was added together with TAT-Cx43_266–283 (Figure 4D) the number of NeuN-positive cells was not reduced when compared with the control (Figure 4A). Quantification of this data (Figure 4E) confirmed the preventive effect of TAT-Cx43_266–283 on the reduction of NeuN-positive cells promoted by KA treatment in neuron-astrocyte co-cultures. It should be mentioned that TAT-Cx43_266–283 did not affect NeuN staining in neuron-astrocyte co-cultures in the absence of KA (Supplementary Figure S1). In agreement with previous studies (Meloni et al., 2014), TAT also showed a slight neuroprotective effect, although it was lower to that found with TAT-Cx43_266–283, suggesting that the region 266–283 in Cx43 has a neuroprotective effect.

To address whether a change in the production of diffusible molecule was involved in the neuroprotective effect, non-contact neuron-astrocyte co-cultures were used. Astrocytes cultured within inserts were placed into wells containing neuronal cultures, as illustrated in Figure 5A. After different treatments, astrocyte inserts were removed and neuronal viability was estimated by MTT viability assays as described in “Materials and Methods” section. In parallel, neurons were subjected to the same treatments in the absence of astrocyte inserts. Consistent with the images shown in Figure 3, in the absence of glial cells TAT-Cx43_266–283 did not significantly modify neuronal death promoted by KA (Figure 5B). However, in the presence of inserts containing astrocytes, TAT-Cx43_266–283 prevented neuronal death promoted by KA (Figure 5B), suggesting that soluble factors released by astrocytes are contributing to this effect.

It has been extensively documented that astrocyte hemichannels when activated release several factors, such as glutamate and ATP that cause neuronal death (Orellana et al., 2011a). Therefore, we explored the participation of these channels in KA-induced neurotoxicity_. Adding 200 μM oATP (P2X receptor blocker), 100 μM BBG (P2X7 receptor blocker) or 20 μM CPP (NMDA receptor blocker) during KA incubation prevented the reduction in NeuN-positive cells caused by KA in neuron-astrocyte co-cultures (Figure 6). Together these data confirmed that inhibitors of ATP- and NMDA-receptors prevented neuronal death in co-cultures treated with KA, suggesting the participation of glutamate and ATP release associated with astrocyte hemichannel activity in this neurotoxic process.

Because astrocyte hemichannels appeared to be involved in KA-mediated neuronal death (Figure 6), we hypothesized that TAT-Cx43_266–283 could prevent hemichannel activation with the subsequent neuroprotective effect. To address this point, first we analyzed the effect of TAT-Cx43_266–283 on glial hemichannel activity by EtBr uptake (Figure 7). Under resting conditions, glial hemichannel activity is negligible (Figure 7B, dotted line), therefore astrocytes were exposed to LPS for 24 h to increase hemichannel activity (Retamal et al., 2007; Orellana et al., 2011a). As shown in Figure 7A, LPS-treated astrocytes showed a strong uptake of the hemichannel permeable tracer. This effect was suppressed by CBX, a hemichannel inhibitor, applied in the presence of LPS, 15 min before and during the EtBr uptake (Figures 7A,B), demonstrating that EtBr uptake occurred through hemichannels. Interestingly, TAT-Cx43_266–283 applied in the same way, i.e., in the presence of LPS, 15 min before and during the EtBr uptake, strongly reduced the activation of hemichannels promoted by LPS (Figures 7A,B). Since TAT-Cx43_266–283 is an inhibitor of c-Src (Gangoso et al., 2014), which is an important mediator of neuroinflammation (Paul et al., 2001; Tai et al., 2013), we investigated the effect of dasatinib, a potent inhibitor of Src, on hemichannel activity. Dasatinib inhibited astrocyte hemichannel activity promoted by LPS (Figure 7B), revealing a role of c-Src in the regulation of hemichannel activity. Furthermore, the results obtained with dasatinib were very similar to those obtained with TAT-Cx43_266–283 suggesting that TAT-Cx43_266–283, by inhibiting c-Src, reduced hemichannel activity in astrocytes. We also analyzed the effect of TAT-Cx43_266–283 on the communication through gap junctions in astrocytes (Figure 7C). The scrape loading assay showed that TAT-Cx43_266–283 did not significantly modify gap junction intercellular communication between astrocytes.

To test the involvement of hemichannel activity on the neuroprotective effect of TAT-Cx43_266–283 in KA-induced neurotoxicity, EtBr uptake was analyzed in neuron-astrocyte co-cultures treated with KA. Figures 7D,E show that KA strongly increased hemichannel activity in these co-cultures. Interestingly, TAT-Cx43_266–283 applied in the presence of KA strongly inhibited hemichannel activity (Figures 7D,E). In fact, the inhibition promoted by TAT-Cx43_266–283 was similar to that promoted by CBX, a hemichannel inhibitor, or by dasatinib, a c-Src inhibitor (Figures 7D,E). These data suggest that TAT-Cx43_266–283 by inhibiting c-Src activity prevented hemichannel activity promoted by KA in neuron-astrocyte co-culture.

We then investigated whether KA, per se, affected astrocyte hemichannel activity. Figure 7B shows that KA did not activate hemichannels in astrocytes. However, when astrocytes were exposed to conditioned medium from KA-treated neuron-astrocyte co-cultures, the activity of astrocyte hemichannels strongly increased (Figure 7B). These results confirm that some factors released by damaged neurons, but not KA, activate astrocyte hemichannels. Altogether, these data suggest that TAT-Cx43_266–283, by inhibiting c-Src, prevents the activation of hemichannels caused by factors released by KA-damaged neurons (Figure 8).