Connexin43 Mimetic Peptides Reduce Swelling, Astrogliosis, and Neuronal Cell Death after Spinal Cord Injury

Abstract

Connexin43 (Cx43) is up-regulated after spinal cord injury (SCI). The authors tested whether mimetic peptides, corresponding to short sequences of rat Cx43, would reduce the severity of damage in a rodent ex vivo model of SCI. Eleven peptides (peptides 1 to 11) corresponding to short amino acid sequences of the extracellular loops of rat Cx43 were tested. Two of these peptides, peptide4 (corresponding to Gap27) and peptide5, significantly reduced the degree of swelling after SCI in this model. Peptide5 produced the more significant reduction in swelling and was analyzed further. Treatment with peptide5 reduced both the level of Cx43 and the number of glial fibrillary acidic protein (GFAP)-positive astrocytes, and at the same time reduced the loss of NeuN-and SMI-32–positive neurons in a concentration-and time-dependent manner. In cell culture, low concentrations of peptide5 prevented hemichannel opening, but did not disrupt gap junctional communication. Higher concentrations prevented hemichannel opening, but also uncoupled existing gap junctions. This study supports the idea that regulation of Cx43 hemichannel opening using mimetic peptides may be a useful treatment for reducing the spread of damage after SCI.

Keywords:

• Connexin43
• hemichannel
• mimetic peptide
• spinal cord injury

INTRODUCTION

Spinal cord injury (SCI) is caused by a mechanical disruption of the spinal cord. Within 1 to 2 h of the injury, there is an acute phase that involves hemorrhage, inflammation, and edema. Cytotoxic edema (intracellular swelling) occurs in astrocytes and is likely due to the presence of a number of molecules and ions in the extracellular space shortly after injury, including glutamate, lactate, K⁺, nitric oxide, arachidonate, reactive oxygen species (ROS), and ammonia (Norenberg et al. 2004).

Connexin (Cx) 43 is the major component of astrocytic gap junctions in the central nervous system (CNS) (Naus et al. 1991). Under physiological conditions, astrocytes play an important role in supporting neurons, forming a glial network and providing neurotrophic factors (Dermietzel and Spray 1998). Under pathological conditions, it is believed that this glial network provides ‘spatial buffering’ that regulates the extracellular K⁺ concentrations and provides energy substrates to neurons (Hansson et al. 2000). After injury, however, a glial scar can form around the site of injury, and this can be detrimental to recovery of the tissue (David and Lacroix 2003). Gap junction changes have been demonstrated in Huntington's disease (Vis et al. 1998), brain tumors (Naus et al. 1991), epilepsy (Fonseca et al. 2002), and brain ischemia (Budd and Lipton 1998; Contreras et al. 2004; Oguro et al. 2001). However, it remains unclear what role gap junctions play and whether they are protective or detrimental (Contreras et al. 2004; Farahani et al. 2005). In a rodent model of stroke, the gap junction inhibitor octanol was reported to reduce the infarct size (Rawanduzy et al. 1997), whereas the treatment of rat brain slices with antisense oligonucleotides, designed against a number of different connexins, reduced neuronal cell death 48 h after an episode of hypoxia and glucose deprivation (Frantseva et al. 2002a, 2002b). Cx43 expression increases in reactive, proliferating astrocytes following ischemic injury, which can result in glial scar formation (Haupt et al. 2007). It has also been demonstrated that mice that are null for the expression of certain connexins in fact have a greater injury after global ischemia (Contreras et al. 2004), and mice that are heterozygous null for Cx43 were reportedly more prone to an increase in infarct size after ischemia (Nakase et al. 2003; Siushansian et al. 2001). Mice deficient for Cx32 show an increased sensitivity to global ischemia in the hippocampal CA1 region (Oguro et al. 2001).

Cx43 expression increases during SCI in the rodent (Lee et al. 2005; Theriault et al. 1997). Levels of Cx43 mRNA and protein increase within 4 h and reach three times the normal level at 4 weeks post injury, whereas no change in expression of Cx32 or Cx36 was reported (Lee et al. 2005). The authors suggested that this increase may enhance local glial networks involved in tissue responses to injury. Theriault et al. reported that Cx43 immunoreactivity is highest in areas associated with neuronal loss whereas Cx43 labeling is normal in areas where neurons are preserved. They demonstrated gap junction disassembly in astrocytes directly adjacent to the injury site, which may be an attempt by astrocytes to prevent communication with neighboring cells, restricting the flow of death signals between these cells (Theriault et al. 1997).

Hemichannels, unopposed connexons in the cell membrane, have been shown to form in a number of cell types (Contreras et al. 2004), including astrocytes (Contreras et al. 2002; Hofer and Dermietzel 1998; Stout et al. 2002). Hemichannels have been demonstrated to have physiological roles such as mediating NAD⁺ release from fibroblasts (Bruzzone et al. 2001), and the control of cell proliferation via autocrine/paracrine NAD⁺-mediated Ca^{2 +} signaling in astrocytes (Verderio et al. 2001). Cx43 hemichannels have also been reported in pathological processes such as cell injury during ischemia (Contreras et al. 2002; Kondo et al. 2000). Cx43 hemichannels are also thought to be involved in Ca^{2 +} signaling via ATP, released through hemichannels, acting on P2Y purinergic receptors on adjacent cells, raising intracellular Ca^{2 +}, and leading to propogation of a Ca^{2 +} wave from cell to cell (Goodenough and Paul 2003) and spreading of a cell death signal from one cell to the next (Krutovskikh et al. 2002). Glutamate has been shown to be released through hemichannels (Ye et al. 2003) and this may play a role in the spread of death signals, as glutamate is known to kill susceptible neurons when released in an uncontrolled manner (Zipfel et al. 2000). Glutamate is released from astrocytes after SCI (Kimelberg 1992). Hemichannels may also be involved in cell volume regulation, as an open hemichannel would allow water influx into the cell (Rodriguez-Sinovas et al. 2007). Cell swelling is a key factor in a number of different insults, including ischemia (Jennings et al. 1990) and spinal cord injury (Norenberg et al. 2004).

Connexin gap junctions and hemichannels can be regulated by various molecules such as heptanol, octanol, and 18α-glycryrrhetinic acid (18α-GCA). These compounds are not considered to be very selective as they can modify other membrane channels, in the case of heptanol and octanol, or act on gap junctions indirectly through a number of signaling pathways, as is the case with 18α-GCA (Evans and Boitano 2001). Antibodies designed against the extracellular region of the connexin molecules, which could interfere with the interaction of two connexin molecules in opposing membranes, have also been investigated as a means to control gap junctions, but with limited success (Meyer et al. 1992). An alternative approach uses connexin mimetic peptides. These are small peptide sequences designed against the extracellular regions of the connexin molecule. They may impair the interactions of the extracellular loops by binding to recognition sites on the hemichannel (Berthoud et al. 2000). Mimetic peptides have been reported to inhibit both gap junction and hemichannel signalling in a number of different in vitro models (Boitano and Evans 2000; Braet et al. 2003; De Vuyst et al. 2007; Kwak and Jongsma 1999; Martin et al. 2005). Connexin mimetic peptides have also been shown to regulate both hemichannels and gap junctions independently of each other, with short incubation times preventing hemichannel opening without affecting gap junction communication, whereas longer incubations can interfere with gap junction communication (Leybaert et al. 2003).

To date, no study has been performed to investigate the effect of regulating Cx43 during spinal cord injury. In this study we set out to determine the effect of treatment using connexin mimetic peptides on the spread of the injury and neuronal cell death after SCI using a rodent ex vivo model.

METHODS

Chemicals and Reagents

Carbenoxolone, 18α-glycyrrhetinic acid, LaCl₃, Triton X-100, Tween 20, and DMSO were obtained from Sigma. All other chemicals were obtained from BDH, unless otherwise stated.

Peptide Design and Manufacture

Overlapping peptide sequences were made against the extracellular regions of rat connexin43 (Table 1 and Figure 1). Overlapping peptides were made so as to cover the entire length of the extracellular loops. Short peptides (11 to 16 amino acids) were used as these have been demonstrated to be more effective than longer peptides (Evans and Boitano 2001). Where possible, amino acids from the membrane-spanning regions were included, as these have been demonstrated to enhance the activity of the peptides (Evans and Boitano 2001). The peptides were supplied by Auspep, Australia. Peptides were synthesised by solid phase using Fmoc chemistries on a Protein Technologies, Symphony instrument. Peptides were purified by high-performance liquid chromatography (HPLC) and the structures confirmed by analytical HPLC and mass spectral analysis.

Figure 1 (A) Schematic of a connexin molecule showing the N-terminal region (NT), the membrane spanning domains (M1 to M4), the extracellular loops (EL1 and EL2), the intracellular loop (IL), and the C-terminal region (CT). (B) Schematic showing the mimetic peptide sequences (see Table 1) and the amino acids of Cx43 to which they correspond. Amino acids that are part of the membrane-spanning domain are underlined and amino acids that are part of the extracellular loop are in italics.

TABLE 1 Sequences of connexin mimetic peptides

Peptidc	Sequence	Reference
1	FEVAFLLIQWYI
2	QWYIYOFSLEAVY
3	SLSAVTTCKRDPCPHQ	Kwak and Jongsma 1999
4 (Gap 27)	SRPTEKTLFII	Chaytor et al. 1997
5	VDCFLSRPTEKT
6	CKRDPCPHQTVCF
7	LGTAVESAWGEDQ
S	CDEQSAFRCNTQQ
9	XTQQPCCENTVCYD
10 (Gap 26)	VCYDKSFPISHVR	Boitano and Evans 2000
11	KSFP1SHVRFWVLQ

Note. Eleven overlapping peptides were designed that span both of the extracellular loops of rat Cx43, and where possible, included areas of the membrane spanning domains. The membrane spanning domains are underlined.

Ex Vivo Spinal Cord Culture

Seven-day-old Wistar rat pups were killed and the spinal cord carefully removed. The cord was washed in Hank's balanced salt solution (HBSS) (Invitrogen) and peripheral nerves removed. The cord was cut into 5-mm segments using a tissue chopper. For the initial experiment, the spinal cord segments were incubated in Neurobasal medium containing 2% B27 supplement, 1 mM l-glutamine (Invitrogen), and 0.45% d-glucose (Sigma) plus the different mimetic peptides, or vehicle, for 4 days in 24-well dishes, with the medium being changed every 24 h. For subsequent experiments using peptide5, spinal cord segments were incubated in medium (as described above) with different concentrations of peptide5 for 1 or 4 days, with medium replaced every 24 h.

Analysis of Tissue Swelling

Spinal cords were excised from 10 Wistar rat pups and cord segments were cultured as described above. Segments were fixed in 4% paraformaldehyde (PFA) for 16 h at 4°C, then washed well with PBS. Tissue swelling was visualized using a Leica M26 stereomicroscope and images captured using a Digital Sight CCD camera (Nikon) and ACT2A software. The amount of swelling was determined using ImageJ software, by measuring the area (viewed from above) of tissue that had exuded beyond the end of the dura divided by the area of the tissue still within the dura, and expressed as a percentage. The degree of swelling of vehicle and control treated cords was converted to 100% and the percentage of swelling of treated cords was expressed relative to this.

Protein Isolation and Western Blot

Spinal cord tissue was homogenised in fresh extraction buffer containing 150 mM sucrose, 15 mM HEPES pH 7.9, 60 mM KCl, 5 mM EDTA pH 8.0, 1 mM EGTA pH 8.0 and 1× Complete protease inhibitor cocktail (Roche). Triton X-100 was added to a final concentration of 1% and incubated on ice for 1 h. The total tissue lysate was centrifuged at 14,000 rpm for 5 min at 4°C. The supernatant was transferred to a fresh tube and the protein concentration determined using the D_C Protein Assay (BioRad).

Equal amounts of protein were separated on 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gels and then transferred to polyvinylidene difluoride (PVDF) membranes and blocked for 30 min with 10% nonfat milk powder in Tris-buffered saline with 0.1% Tween 20 (TBS-T) at room temperature. Immunodetection was carried out using the following antibodies: mouse anti-Cx43 (M3067, Chemicon), rabbit anti-glial fibrillary acidic protein (GFAP) (Z0334, Dako), mouse anti-Neuronal N (NeuN) (MAB377, Chemicon), and mouse anti-nonphosphorylated neurofilament H (SMI-32) (Sternberger Monoclonals). Antibodies were diluted with 5% bovine serum albumin (BSA) in TBS-T and blot probed overnight at 4°C. The blot was washed as above and then incubated with a 1:2000 dilution of either anti-mouse or anti-rabbit antibody conjugated with horseradish peroxidase (Sigma) in 5% nonfat milk powder in TBS-T for 2 h at room temperature. The blot was washed as before and immunoreactive bands were visualized using ECLplus (enhanced chemiluminescence) detection reagent (GE Healthcare).

Tissue Processing and Immunohistochemistry

Spinal cord segments were fixed in Bouins fixative (75% picric acid, 10% formaldehyde, 5% glacial acetic acid) for 16 h at 4°C. Tissue was washed well with PBS and then paraffin embedded. Longitudinal, 8 μ m thick sections were cut from corresponding areas within each sample and mounted on glass slides. Prior to immunohistochemistry, wax was removed by heating slides to 64°C for 20 min, followed by xylol/alcohol washes. Slides were rinsed in water and washed three times in PBS. Antigen retrieval was performed using Dako Target Retrieval Solution (Dako) at 98°C for 30 min. Slides were then washed in PBS.

Sections were blocked using 2% normal goat serum (NGS) in PBS with 0.2% Triton X100 (PBS-T) for 2 h at room temperature. Immunohistochemistry was carried out using the primary antibodies described above. Antibodies were diluted with 2% NGS in PBS-T and incubated overnight at 4°C. Slides were washed in PBS and incubated with a 1:400 dilution of anti-rabbit Alexa-568, anti-mouse Alexa-568, or anti-mouse Alexa-488 (Invitrogen) in 2% NGS in PBS-T for 3 h at room temperature. Slides were washed with PBS, coverslip mounted using ProLong Gold Antifade Reagent (Invitrogen). Staining was visualized using a Leica DMR fluoresence microscope and images captured using a Digital Sight CCD camera (Nikon) and EclipseNet software.

In Vitro Peptide Efficacy Analysis: Low Ca²⁺ Dye Influx Assay

Cells were plated at a density of 1 × 10⁴ NT2/D1 cells/well into 96-well plates in 10% fetal bovine serum (FBS) Dulbecco's modified Eagle's medium (DMEM/F12) containing 2 mM l-glutamine at 37°C, 5% CO₂. The next day cells were rinsed twice with Ca^{2 +}-and Mg^{2 +}-free HBSS containing 25 mM HEPES pH 7.4 and 1 mM EGTA. Cells were then incubated for 30 mins in the same solution containing 2 mM propidium iodide (Invitrogen) and either vehicle (DMSO for carbenoxolone, H₂O for all other compounds), 100 μ M carbenoxolone, 5 or 500 μ M peptide5, 500 μ M peptide8, or 200 μ M LaCl₃. For the LaCl₃ experiments only 100 μ M EGTA was used, as EGTA has a high affinity for La^{3 +} and therefore forms a complex (Braet et al. 2003). For each culture, four images were visualized using a Nikon TE2000E inverted fluorescent microscope and captured using a Digital Sight CCD camera (Nikon) and EclipseNet software and the number of cells that had taken up propidium iodide was counted using ImageJ software.

In Vitro Peptide Efficacy Analysis: Cell Settlement Assay

Cells were plated at a density of 6.5 × 10⁴ NT2/D1 cells/well into 24-well plates in 10% FBS DMEM/F12 containing 2 mM l-glutamine at 37°C, 5% CO₂. The next day, half the wells were incubated with medium containing 10 μ g/ml Hoechst 33342 and 4 μ M calcein-AM (Molecular Probes) for 30 min. After this time, the cells were washed 3 times for 5 min with PBS and trypsinized with 0.05% trypsin-EDTA. Cells were resuspended in 1 ml medium and 50 μ l of the cell suspension was added to a confluent monolayer of unlabelled cells. After 30 min, vehicle (DMSO for carbenoxolone, H₂O for all other compounds), 5 or 500 μ M peptide5, 500 μ M peptide8, or 100 μ M carbenoxolone was added to the wells and incubated for a further hour. For each well, six images were visualized on a Nikon TE2000E inverted fluorescent microscope and captured using a Digital Sight CCD camera (Nikon) and EclipseNet software and the number of cells that had taken up calcein dye from a Hoechst-labeled feeder cell were counted.

Statistical Analysis

Statistical analysis was carried by one-way analysis of variance (ANOVA) using Prism 4.0 software. If significant (p < 0.05), a Dunnet's multiple comparision test was performed.

RESULTS

Connexin Mimetic Peptides Reduce Swelling in Ex Vivo Spinal Cord Cultures

In our ex vivo model of spinal cord injury, tissue swells within 24 h of the spinal cord segments being placed in culture with the tissue extruding out of the end of the dura (Figure 2A, panel b, arrows). Tissue swelling is one of the first events to occur after spinal cord injury in vivo. In order to determine if the use of mimetic peptides to regulate Cx43 gap junctions or hemichannels could prevent this swelling after spinal cord injury, ex vivo spinal cord cultures where cultured in the presence of 500 μ M of each peptide designed against the extracellular regions of rat Cx43 (Table 1) or vehicle for 24 h. The tissue was fixed and the degree of swelling was measured and used to assess the degree of damage. Both peptide4 (Gap27) and peptide5 reduced swelling compared with controls (Figure 2B), with peptide5 reducing swelling by nearly 50%. As can be seen in Figure 2C, in vehicle-treated cultures the tissue protrudes out of the ends of the dura (arrow), whereas the dura appears to retract and looks damaged. In peptide-treated cultures, however, the swelling is markedly reduced (Figure 2C, panel b), and the dura does not retract and remains in much better condition. As peptide5 lead to a greater reduction in swelling, this peptide was used for all further studies.

Figure 2 Culture of spinal cord segments leads to swelling which is reduced by connexin mimetic peptides. (A) Photograph showing the increase in swelling out of the ends of the dura (arrows) in ex vivo spinal cord segments after 24 h in culture (panel b), compared with segments fixed at time 0 (panel a). (B) Graph demonstrating the degree of swelling at 24 h, after treatment with connexin mimetic peptides (1 to 11) or vehicle. Peptide4 and peptide5 show a significant reduction in swelling; * p < 0.05; ** p < 0.01. (C) Photograph showing the difference in swelling in ex vivo spinal cord segments after 24 h culture with vehicle (panel a) and 500 μ M peptide5 (panel b).

The Effect of Peptide5 Is Concentration and Time Dependent

Ex vivo cultures were maintained for 24 h in the presence of different concentrations of peptide5, after which they were fixed and the swelling assessed. Peptide8 was used as an inactive control peptide, as this did not reduce the degree of swelling (Figure 2B). As demonstrated in Figure 3, concentrations of peptide5 between 5 and 500 μ M significantly reduced the degree of swelling. We next determined whether a longer incubation would improve the effect of peptide5 on reducing swelling. When peptide5 was present for the entire length of the experiment (4 days), only concentrations below 100 μ M reduced swelling, with 250 and 500 μ M peptide5 having no effect on swelling compared with the control peptide (Figure 4).

Figure 3 Concentration-dependent reduction of spinal cord swelling after treatment with peptide5 for 24 h. Spinal cord swelling was measured after treatment with different concentrations of peptide5 and inactive control peptide (peptide8) for 24 h and expressed as a percentage of the swelling seen in vehicle treated cords. All concentrations of peptide5, 5 μ M and above show a significant reduction in swelling; *** p < 0.001.

Figure 4 Concentration-dependent reduction of spinal cord swelling after treatment with peptide5 for 4 days. Spinal cord swelling was measured after treatment with different concentrations of peptide5 and inactive control peptide (peptide8) for 4 days and expressed as a percentage of the swelling seen in vehicle-treated cords. Concentrations of peptide5 between 5 and 100 μ M show a significant reduction in swelling; * p < 0.05; *** p < 0.001.

Peptide5 Prevents Astrogliosis after SCI

Astrocytes have been shown to be involved in swelling or edema after SCI. As peptide5 was able to reduce swelling in our cultures, we then wanted to determine whether it was having an effect on the number of astrocytes in our cultures. Spinal cord segments were cultured for up to 4 days, with 5 μ M peptide5 or peptide8 present for the first 24 h. Western blotting and immunohistochemistry was then carried out for GFAP, a marker for activated astrocytes. As demonstrated in Figure 5A, in control-treated cultures, the levels of GFAP are increased at day 1 and are maximal at day 3 before being reduced again by day 4. In contrast, a much smaller increase in GFAP expression is seen with peptide5 treatment at day 1 before levels reduce to those of day 0 cultures. For immunohistochemistry, 8-μ m sections were cut from corresponding regions of the spinal cord to allow for comparison. Immunohistochemistry demonstrated that in peptide5-treated cultures, there is decreased GFAP staining, compared with control cultures at both 1 and 4 days of culture (Figure 5B).

Figure 5 Peptide5 reduces GFAP protein levels in ex vivo spinal cord segments after injury. (A) Western blot of GFAP protein levels in ex vivo spinal cord segments treated with 5 μ M peptide5 or control peptide (peptide8) for 24 h and cultured for up to 4 days, showing decreased GFAP protein levels in the presence of peptide5. (B) Immunohistochemical staining of corresponding sections from ex vivo spinal cord cultures treated as above. Scale bar = 100 μ m.

Peptide5 Reduces Cx43 Protein Levels after SCI

As Cx43 is the major connexin expressed in astrocytes, we sought to determine what role peptide5 might have on Cx43 protein levels. Spinal cord segments were cultured at various timepoints and protein lysates prepared. As seen in Figure 6, Cx43 protein levels increase at 4 h after injury. From 8 h, it reduces over 4 days but remains higher than that seen in day 0 cultures. In spinal cord segments cultured in the presence of 5 μ M peptide5, there is also an increase in the level of Cx43 by 4 h, but this is markedly reduced compared to control (peptide8). Cx43 levels then reduces again until day 4.

Figure 6 Peptide5 reduces Cx43 protein levels in ex vivo spinal cord segments after injury. Ex vivo spinal cord segments treated with 5 μ M peptide5 or control peptide (peptide8) for up to 24 h and cultured for up to 4 days. Western blot analysis shows the increase in Cx43 protein seen in control (peptide8)-treated cultures is reduced in peptide5-treated cultures.

Peptide5 Reduces Neuronal Cell Death after SCI

We next wanted to determine if peptide5 would lead to an increased survival of neurons. Spinal cords were cultured for up to 4 days, then Western blotting and immunohistochemistry was carried out for the neuronal markers NeuronalN (NeuN) and neurofiliment H (SMI-32). NeuN is a marker of all mature neurons and SMI 32 visualizes neuronal cell bodies, dendrites, and some thick axons in the central and peripheral nervous systems. As demonstrated by Western blot, in control-treated slides the levels of both NeuN (Figure 7A) and SMI-32 (Figure 8A) decrease over the 4 days of the culture. Treatment with peptide5 reduces this loss of the neuronal markers NeuN and SMI-32 labeling. This was confirmed by immunohistochemistry. In the presence of peptide5, there is a reduction in the loss of NeuN labeling and there are more large nuclei present than in control treated slides at 4 days (Figure 7B). A similar pattern is seen with immunohistochemistry for SMI-32 (Figure 8B). After 4 days' culture, there are still SMI-32–positive neurons present in peptide5-treated cultures, although this is reduced from day 1. In control-treated slides, there are no SMI-32–positive cells remaining, although staining is still present in the white matter (Figure 8B).

Figure 7 Peptide5 prevents the reduction of NeuN protein levels in ex vivo spinal cord segments. (A) Western blot of NeuN protein levels in ex vivo spinal cord segments treated with 5 μ M peptide5 or control peptide (peptide8) for 24 h and cultured for up to 4 days, showing increased neuronal survival in the presence of peptide5. (B) Immunohistochemical staining of corresponding sections for NeuN at day 1 and day 4 of culture. Scale bar = 100 μ m.

Figure 8 Peptide5 prevents the reduction of SMI-32 protein levels in ex vivo spinal cord segments. (A) Western blot of SMI-32 protein levels in ex vivo spinal cord segments treated with 5 μ M peptide5 or control peptide (peptide8) for 24 h and cultured for up to 4 days, showing increased neuronal survival in the presence of peptide5. (B) Immunohistochemical staining of corresponding sections for SMI-32 at day 1 and day 4 of culture. Scale bar = 100 μ m.

Peptide5 Regulates Hemichannels and Gap Junctions in a Concentration-Dependent Manner

Previous studies using mimetic peptides to regulate gap junctions used considerably higher concentrations than those found to be most effective in this study (De Vuyst et al. 2007; Kwak and Jongsma 1999; Matchkov et al. 2006). It has also been demonstrated that mimetic peptides can regulate both hemichannels and gap junctions independently of each other, with short incubation times regulating hemichannels alone, and longer incubation periods being necessary to inhibit gap junction communication. Because in this study we found that a low (5 μM) concentration of peptide was as effective as higher doses and that a longer incubation period was ineffective compared with shorter incubations, we investigated whether our mimetic peptide was regulating hemichannels or gap junction communication or both. Firstly, in order to determine if peptide5 could regulate hemichannels, low Ca^{2 +} experiments were carried out using the NT2/D1 cell line, which expresses high levels of Cx43. There was a significant uptake of propidium iodide into the vehicle-treated cells (Figure 9A and Figure 9B). In contrast, the hemichannel blockers carbenoxolone (100 μ M) and LaCl₃ (200 μ M) significantly reduced dye uptake (Figure 9A and Figure 9B). In the presence of 5 and 500 μM peptide5, dye uptake was reduced to the levels seen with carbenoxolone and LaCl₃ (Figure 9A and Figure 9B). Incubation with 500 μ M control peptide8 had no effect on propidium iodide uptake compared with vehicle-treated cells.

Figure 9 Peptide5 prevents hemichannel opening. (A) Fluorescent images of NT2/D1 cells incubated with propidium iodide in low Ca^{2 +} conditions. Both 5 and 500 μ M peptide5 reduce the uptake of propidium iodide to a similar degree as LaCl₃ and carbenoxolone while peptide8 has no effect on dye uptake. (B) Graph demonstrating that the number of cells taking up dye is significantly reduced in the presence of peptide5, LaCl3 and carbenoxolone but not peptide 8; **p < 0.01.

To determine the effect of peptide5 on gap junction communication, we used a cell settlement assay. When dye loaded NT2/D1 cells were replated onto a population of unloaded cells and imaged 1 h later, the spread of calcein dye into the unlabelled cells was seen (Figure 10A and Figure 10B). In the presence of the gap junction blocker carbenoxolone (100 μ M), however, the spread of dye is significantly reduced (Figure 10A and B). When the loaded cells were replated in the presence of peptide5, dye spread is blocked at a concentration of 500 μ M but not at 5 μM (Figure 10A and Figure 10B). Incubation with 500 μ M control peptide8 had no effect on calcein dye spread compared with vehicle-treated cells.

Figure 10 High concentrations of peptide5 prevents gap junctional communication. (A) Fluorescent images of a cell settlement assay using NT2/D1 cells. A concentration of 500 μ M peptide5 reduces the transfer of calcein dye from loaded to unloaded cells to a similar degree as carbenoxolone-treated cells, whereas 5 μ M peptide5 and 500 μ M peptide8 has no effect on dye transfer. (B) Graph demonstrating the number of cells to which dye is transferred is significantly reduced in the presence of 500 μ M peptide5 and carbenoxolone but not 5 μ M peptide5 or 500 μ M peptide8; *p < 0.05.

DISCUSSION

In this study, we demonstrated that connexin mimetic peptides prevented swelling, reduced astrogliosis, and increased neuronal survival after spinal cord injury. We also demonstrated that this protective effect was concentration and time dependent. When spinal cords were treated with doses of peptide5 between 5 and 500 μ M for 24 h, swelling was significantly reduced. However, when the cultures were treated with peptide5 for 4 days, only concentrations below 100 μ M had an effect on swelling. This suggests that Cx43 may play two different roles during spinal cord injury, being involved in the spread of damage in the early stages after injury, but at later stages it is involved in glial responses to injury and to allow spatial buffering by astrocytes. It also suggests that the peptide is acting via different mechanisms. In tissue culture experiments, we demonstrated that a low concentration (5 μ M) of peptide5 was able to prevent hemichannel opening but had no effect on gap junction communication, whereas high concentrations (500 μ M) prevented both hemichannel opening and gap junctional communication. We suggest that low concentrations of peptide5 prevent hemichannel opening early after injury (within the first 24 h) and prevent the spread of neurotoxic molecules. At later time points (after 24 h), high concentrations of peptide5 inhibit gap junctional communication and interfere with the spatial buffering activity of astrocytes, whereas low concentrations would be unable to do this. It is believed that mimetic peptides diffuse into the intercellular cleft where they are able to bind hemichannels and prevent their assembly into gap junctions, or block existing gap junctions (Evans et al. 2006). High concentrations of peptide may be effective, whereas low concentrations are insufficient to achieve this. On the other hand, low concentrations would be sufficient to interact with hemichannels, which exist unopposed on the cell membrane. As low concentrations of peptide5 do not have a detrimental effect on swelling when in the culture for extended periods, this suggests that hemichannels are not involved in the protective effects of gap junctions at the later stage.

There is evidence to suggest that Cx43 hemichannels may be involved early during SCI. After SCI, there is an acute phase that begins 1 to 2 h after injury. This involves hemorrhage, inflammation, and edema. Cytotoxic edema (intracellular swelling) occurs in astrocytes and is likely due to the presence of a number of molecules in the extracellular space shortly after injury, including glutamate, lactate, K⁺, nitric oxide, arachidonate, ROS, and ammonia (Kimelberg 1992; Norenberg et al. 2004), many of which can pass through or regulate hemichannels. Glutamate has been reported to pass through hemichannels (Ye et al. 2003), whereas oxidative stress and nitric oxide have been demonstrated to increase Cx43 hemichannel opening in astrocytes (Retamal et al. 2006). It has been suggested that the coupling of astrocytes may be important in determining the susceptibility of neurons to oxidative stress (Perez Velazquez et al. 2006). Treatment of cortical astrocytes, subjected to metabolic inhibition, with lipoxygenase inhibitors reduces hemichannel opening (Contreras et al. 2002), and arachadonic acid potentiates hemichannel opening (De Vuyst et al. 2007), suggesting arachadonic acid metabolites may play a role in hemichannel regulation. Increasing extracellular K⁺ concentrations have been reported to increase the opening for Cx50 and Cx46 hemichannels (Srinivas et al. 2006), but it is unknown if extracellular K⁺ opens Cx43 hemichannels.

There are conflicting data on the role of Cx43 expression and signaling after brain injury, with a number of studies indicating that increased Cx43 levels are detrimental (Frantseva et al. 2002a; Frantseva et al. 2002b) and others reporting that a lack of Cx43 leads to an increase in the size of the injury (Nakase et al. 2004; Siushansian et al. 2001). The studies where Cx43 expression appeared to be detrimental used slice culture and antisense oligonucleotides (AS) or gap junction blockers to down-regulate connexin expression. As AS treatment only down-regulates connexin expression transiently, this would prevent connexin signaling during the early stages of injury, but could allow recovery of connexin expression and gap junctional communication (Green et al. 2001; Qiu et al. 2003). Studies that report a loss of connexins as detrimental used models in which connexin expression was prevented by gene knockout. This would, however, prevent the recovery of gap junction communication and the cell-cell communication that is required for subsequent spatial buffering, neuronal survival and subsequent tissue coupling function. We believe our data supports the idea that connexin signaling is detrimental early after injury, but is required later for recovery of the tissue.

Treatment with 5 μ M peptide5 prevented the increase in GFAP seen within 24 h in our ex vivo cultures. Reactive astrogliosis after spinal cord injury is believed to be involved in providing a protective environment for neurons (Lee et al. 2005; Theriault et al. 1997). This might suggest that prevention of astrocytosis may be detrimental to neuronal protection. However, down-regulation of connexin signaling during the early stages after injury would prevent the release of neurotoxic factors into the extracellular matrix and subsequent neuronal cell death and therefore the signals that lead to the increase in astrocyte number. Although the early stages of astrogliosis may be neuroprotective, astrocytes eventually form a glial scar that inhibits axonal growth and functional recovery. Treatment with 5 μ M peptide5 also reduced the increase in Cx43 levels seen after injury. It is unclear whether this is a direct action of the mimetic peptide, such as internalization and degradation of the connexin, or is an indirect effect of the peptide in mediating the response to injury.

Peptide5 prevented loss of neurons after spinal cord injury. In humans neurons are very vulnerable to damage after SCI, and this occurs within hours of the injury. By preventing edema, peptide5 protects the neurons from necrosis, which would in turn likely prevent the increase in astrogliosis. Therefore, early treatment with compounds such as peptide5 may create an environment for surviving neurons to regenerate axons through the injury site, in the absence of glial scarring.

Interestingly, although peptide4 (Gap27) reduced swelling, it was not as marked as that seen with peptide5. Both Gap27 and peptide5 share the SRPTEKT sequence, whereas Gap27 contains the amino acids LFII from the membrane region and peptide5 contains the extracellular loop amino acids VDCFL. It has been suggested that peptides that contain regions from the membrane domain are more effective in blocking gap junctional communication (Goodenough et al. 1996; Warner et al. 1995). However, this was not seen in the present study.

Warner et al. (1995) used a sequence similar to peptide5, designed against the equivalent extracellular loop region of Cx32, to look at gap junction formation in chick heart myocytes. Although they reported that it was functional in preventing gap junction formation, the effect was not as great as that seen with Gap27 (equivalent to our peptide4) in their studies. However, the chick myocytes were coexpressing connexins 26, 32, and 43. The majority of studies using connexin mimetic peptides have been carried out on monolayers in cell culture. It is possible that there may be issues of peptide penetration into tissue, depending on the amino acid sequence and its hydrophobicity, which could affect its ability to penetrate our spinal cord tissue. It has been reported that there is no difference in the efficacy between the commonly used peptides Gap26 and Gap27 in preventing gap junction communication (Chaytor et al. 1997; Warner et al. 1995). However, although our peptide4 (equivalent to Gap27) did have an effect on swelling, in our hands peptide10 (Gap26) had no effect. The reason for this is unclear but again may be due to the ability of the peptide to penetrate the tissue based on its amino acid sequence.

In conclusion, we have shown that Cx43 channels may have a dual role after spinal cord injury. They appear to be involved in astrocytic swelling and the release of neurotoxins during the early stages of injury, possibly via hemichannels. However, recovery of gap junction expression is likely to be necessary for spatial buffering that is required for long-term survival of neurons.

Louise F. B. Nicholson and Colin R. Green contributed equally to this work.

This work was supported by the Royal Society of New Zealand Marsden Fund and the Catwalk Trust. The authors thank Dr. Hannah Gibbons for her critical reading of the manuscript.